A Word Bank for Academic Writing

A Collection of important words and great expressions.

788 important words and 795 great expressions.


Useful links:
Google, WIKI, COCA, Google Translate, Baidu Translate, Youdao Translate


Learning:


Words

  1. 90th percentile  第90%的数 [?]
  2. ASAP abbr. as soon as possible [?]
  3. Albeit conj. 尽管,虽然 [?]
  4. Augmented Reality  虚拟现实 [?]
  5. Dirac n. 狄拉克; 迪拉克 [?]
  6. Quantile n. 分位数 [?]
  7. Sicily n. 【意大利】西西里岛 [?]
  8. a.k.a. abbr. also known as [?]
  9. abbreviate v. 缩写 [?]
  10. ablation n. n.(冰、雪的)消融;冰消作用;(岩石的)磨蚀 [?]
  11. absence n. 缺席,不在 [?]
  12. abundance n. 大量; 丰盛; 充裕 [?]
  13. accordance n. 一致 [?]
  14. acoustic adj. 声音的; 音响的; 听觉的; 原声的; 自然声的 [?]
  15. acoustic sensor n. 声波传感器 [?]
  16. acquaint adj. 使...熟悉 [?]
  17. acute adj. 十分严重的; 灵敏的 [?]
  18. adjacent adj. 相邻的 [?]
  19. adversarial adj. 对立的,敌对的 [?]
  20. adversarial adj. 对立的;对手的;对抗性的;对抗的 [?]
  21. adverse adj. 不利的; 有害的; 反面的,恶劣的 [?]
  22. adverse adj. 不利的,有害的 [?]
  23. adversely adv. 反而; 不利地;逆地;反对地 [?]
  24. advocate v. 提倡 [?]
  25. advocate v. 提倡,倡导 [?]
  26. aerosol n. 气溶胶   [?]
  27. aforementioned adj. 前面提到的 [?]
  28. agglomerative adj. 附聚的;烧结的;凝聚(结)的 [?]
  29. aggravation n. 加重,恶化 [?]
  30. aggregating adj. 总计的,合计的 [?]
  31. agnostic adj. 不可知的 [?]
  32. airglow n. 辉光 [?]
  33. aleatoric adj. 偶然的;机遇的;任意的 [?]
  34. alleviate v. 减轻; 缓和; 缓解 [?]
  35. alleviate v. 减轻; 缓和; 缓解 [?]
  36. alleviate v. 减轻; 缓和; 缓解 [?]
  37. alleviate v. 减轻,缓和 [?]
  38. allotte v. 分配,配给(时间、钱财等); 分派(任务等) [?]
  39. alternate adj./v. 交替的; 轮流的; 间隔的; 每隔(…天等)的/使交替; 使轮流; 交替; 轮流 [?]
  40. ambient adj. 周围的,周围环境的 [?]
  41. ambiguity n. 模棱两可 [?]
  42. ambiguity n. 模棱两可 [?]
  43. ample adj. 足够的; 丰裕的; 丰满的; 硕大的 [?]
  44. analogous adj. 模拟的 [?]
  45. analogous adj. 相似的,类似的   [?]
  46. analogy n. 类比; 比拟; 比喻; 类推 [?]
  47. analogy n. 类比; 比拟; 比喻; 类推 [?]
  48. ancestry n. 祖宗,祖先 [?]
  49. ancillary adj. 辅助的; 补充的; 附属的; 附加的 [?]
  50. annotate v. 注释 [?]
  51. annotation n. 注释 [?]
  52. anomaly n. 异常事物; 反常现象 [?]
  53. anthropogenic adj. 人为的;人类起源的;人类发生的;人类的 [?]
  54. appealling adj. 有吸引力的 [?]
  55. applicable adj. 适用的; 合适的 [?]
  56. applicable adj. 适用的,合适的 [?]
  57. arguably adv. (常用于形容词比较级或最高级前) 可论证地,按理 [?]
  58. arguably adv. 可论证地 [?]
  59. artifact n. 人工制品 [?]
  60. assertive adj. 坚定自信的; 坚决主张的 [?]
  61. asynchronously adv. 异步地,不同步地 [?]
  62. atlas n. 地图册; 地图集 [?]
  63. attenuation n. 衰减;变薄;稀释 [?]
  64. attest v. 证实; 是…的证据; 作证,证明 [?]
  65. augment v. 增强,提高 [?]
  66. auxiliary adj. 辅助的,备用的 [?]
  67. auxiliary adj. 辅助的,备用的 [?]
  68. avionic n. 航空电子设备 [?]
  69. axon n. 轴突(神经细胞的突起) [?]
  70. backhoe n. 挖掘机 [?]
  71. barometer n. 气压计 [?]
  72. barometer n. 气压计 [?]
  73. bearing n. 轴承 [?]
  74. bearing angle  方位角,象限角 [?]
  75. beset v. 困扰; 威胁 [?]
  76. bias n. 偏差 [?]
  77. bias n. 偏差,差异 [?]
  78. binocular adj. 双目并用的,双眼的 [?]
  79. blueish adj. 带蓝色的,有点蓝的 [?]
  80. blunder n. 错误 [?]
  81. bold adj. 大胆自信的; 敢于冒险的; 明显的; 粗体的; 黑体的 [?]
  82. bombastic adj. 夸夸其谈的; 大话连篇的; 词藻华丽空洞的 [?]
  83. boostrapping n. 引导程序 [?]
  84. breadth n. 宽度,广泛 [?]
  85. breakdown v. 拆分,分类 [?]
  86. brittleness n. 脆性,脆度;脆弱性 [?]
  87. bulldozer n. 推土机 [?]
  88. bump sensor n. 碰撞传感器 [?]
  89. bunker n. 地堡; 掩体; 煤舱 [?]
  90. by leveraging...  通过依靠,使用... [?]
  91. camcorder n. 摄像机 [?]
  92. camouflage n. (军事上的) 伪装,隐蔽 [?]
  93. camouflaged adj. 伪装的,掩饰的 [?]
  94. canonical adj. 标准;规范的;典型 [?]
  95. canonical adj. 经典的, (数学表达式) 最简洁的 [?]
  96. captioning n. 在图片上加说明 [?]
  97. cartography n. 制图学; 地图绘制 [?]
  98. catadioptic adj. 反射的,折射的 [?]
  99. causal adj. 因果关系的; 前因后果的; 原因的   [?]
  100. caveat n. 警告,告诫 [?]
  101. chronological adj. 按发生时间顺序排列的 [?]
  102. chronologically adv. 按时间顺序发生/排列地 [?]
  103. clamp v. 固定,保持不变,夹紧,夹住 [?]
  104. cleanroom n. 无尘室 [?]
  105. cluster v. 集中,成群 [?]
  106. clutter n. 杂乱的东西 [?]
  107. cluttered adj. 杂乱的,挤满的 [?]
  108. coarse-grained  粗颗粒的,粗粒度 [?]
  109. codec n. 编码译码器 [?]
  110. coherence n. 连贯性; 条理性 [?]
  111. coherence n. 连贯性;条理性 [?]
  112. coherent adj. 合乎逻辑的; 有条理的; 清楚易懂的; 有表达能力的 [?]
  113. collision-less  无碰撞 [?]
  114. combustion n. 燃烧; 燃烧过程 [?]
  115. commercial off-the-shelf  现货 [?]
  116. commercialization n. 商业化 [?]
  117. commodity n. 商品,货物,日用品 [?]
  118. compelling adj. 引人入胜的 [?]
  119. compelling adj. 引人入胜的; 令人信服的 [?]
  120. complement v. 补足,补充,使完美   [?]
  121. complementary adj. 互补的; 补充的; 相互补足的 [?]
  122. complication n. 复杂,混乱 [?]
  123. compounded adj. 复合的;合成的 [?]
  124. comprehensive adj. 所有的 [?]
  125. comprise v. 由...组成;占;由……构成 [?]
  126. compromise n. 妥协,危及,损害,连累 [?]
  127. con n. 反对的意见,缺点 [?]
  128. concatenated adj. 连续的,连接的 [?]
  129. concatenated adj. 连锁状的 [?]
  130. conceptually adv. 概念上地,理论上地 [?]
  131. concise adj. 简明的,简练的 [?]
  132. concisely adv. 简明地,简洁地 [?]
  133. concurrently adv. 兼;同时发生地 [?]
  134. condense v. (由气体) 冷凝; (使气体) 凝结; (使) 浓缩,变浓,变稠; 简缩,压缩(文字、信息等) [?]
  135. conducive adj. 使容易(或有可能)发生的 [?]
  136. confine v. 限制; 限定 [?]
  137. confine v. 限制; 限定; 监禁; 禁闭; 使离不开(或受困于床、轮椅等); [?]
  138. confined adj. 狭窄而围起来的 [?]
  139. confounding adj. 使人困惑惊讶的,使惊疑的   [?]
  140. congestion n. 拥塞,塞车 [?]
  141. conjecture v. 猜测; 推测 [?]
  142. consecutive adj. 连续不断的 [?]
  143. consecutive adj. 连续不断的 [?]
  144. consecutive adj. 连续不断的 [?]
  145. consecutive adj. 连续不断的 [?]
  146. consecutive adj. 连续不断的 [?]
  147. consecutive adj. 连续不断的 [?]
  148. consensus n. 一致的意见; 共识 [?]
  149. consequently adv. 因此,所以 [?]
  150. consolidate v. 使加强; 使巩固; (使) 结成一体,合并 [?]
  151. conspicuous adj. 易见的; 明显的; 惹人注意的   [?]
  152. constellation n. 星座(卫星) [?]
  153. constituent adj. 组成的; 构成的 [?]
  154. constituent adj. 组成的; 构成的 [?]
  155. contaminated adj. 受污染的 [?]
  156. contaminated adj. 被污染的 [?]
  157. contiguous adj. 相接的; 相邻的 [?]
  158. contiguous adj. 相邻的,相接的 [?]
  159. contradict v. 反驳; 驳斥; 批驳; 相抵触; 相矛盾; 相反   [?]
  160. converge v. 会聚,收敛 [?]
  161. convergence n. 收敛 [?]
  162. convergence n. 趋同,融合,收敛 [?]
  163. corner stone n. 基石; 房角石; 奠基石; 墙角石; 基础 [?]
  164. cornerstone n. 基石;基础;奠基石;最重要部分 [?]
  165. corollary n. 必然的结果(或结论) [?]
  166. corridor n. (建筑物内的)走廊,过道,通道; (火车上的)走道; 走廊 [?]
  167. corridor n. 过道,走廊 [?]
  168. corroborated adj. 证实,确证(陈述、理论等) [?]
  169. corroborates v. 证实,确证(陈述、理论等) [?]
  170. corruption n. 污损,损坏,腐败 [?]
  171. counterpart n. 对应物,配对物 [?]
  172. course n. (有关某学科的系列) 课程,讲座; (大学中要进行考试或取得资格的) 课程; (船或飞机的) 航向,航线 [?]
  173. covariance n. 协方差 [?]
  174. crawl v. 爬行,爬虫(数据) [?]
  175. criteria n. 标准,尺度 [?]
  176. criterion n. 标准,原则 [?]
  177. crop v. 裁剪,剪短 [?]
  178. crowdsourcing n. 众包;群众外包;众包模式 [?]
  179. cubic adj. 立方的,立方形的 [?]
  180. cull v. 部分捕杀,部分去除/删除 [?]
  181. curse n. 诅咒 [?]
  182. curse n. 诅咒,咒语 [?]
  183. dashed line  虚线 [?]
  184. daunting adj. 使人畏惧的; 令人胆怯的; 让人气馁的 [?]
  185. de-facto  事实上,实际上(没有规定但实际上...) [?]
  186. decibel n. 分贝 [?]
  187. declassification n. 解密; 解密报告; 解密处理 [?]
  188. dedicated adj. 专注的 [?]
  189. deduce v. 推演,推算 [?]
  190. defects n. 缺点; 缺陷; 毛病   [?]
  191. degrade v. 降低,削弱 [?]
  192. delicate adj. 精致的,微妙的 [?]
  193. delimit v. 定…的界限; 限定; 界定 [?]
  194. delineate v. (详细地) 描述,描画,解释 [?]
  195. delineation n. 画轮廓;略图;图形 [?]
  196. delineation n. 轮廓;略图;图形 [?]
  197. demanding adj. 要求高的; 费力的; 要求极严的; 苛求的 [?]
  198. demographic n./adj. (某个地方或社会的) 人口统计数据; (尤指特定年龄段的) 人群/人口学的; 人口统计学的 [?]
  199. dendrite n. 树突(位于神经元末端的细分支) [?]
  200. dendrogram n. 系统树图 [?]
  201. densify v. 使稠密;使增加密度 [?]
  202. depict v. 描写,叙述 [?]
  203. depict v. 描写,画,如...所示 [?]
  204. depict v. 描绘,图示 [?]
  205. deposite v.  存储 [?]
  206. derive v. 推导,得到 [?]
  207. desirable adj. 可取的, 值得拥有的 [?]
  208. determination n. 测定;确定;决心;计算 [?]
  209. deterministic adj. (思想、解释等) 基于决定论的; (力量、因素) 不可抗拒的,不可逆转的 [?]
  210. deterministic adj. 决定性的 [?]
  211. deviate v. 背离; 偏离; 违背 [?]
  212. diameter n. 直径 [?]
  213. diffuse adj. 不清楚的; 难解的; 难以区分的 [?]
  214. digression n. 离题,题外话   [?]
  215. dim adj. 暗淡的; 昏暗的; 微弱的; 不明亮的; 光线暗淡的 [?]
  216. diminish v. 减少; (使) 减弱,缩减; 降低 [?]
  217. diminish v. 减少; (使) 减弱,缩减; 降低 [?]
  218. diminished adj. 减少的,削弱的,降低的 [?]
  219. disarm v. 解除...武装 [?]
  220. disastrous adj. 极糟糕的; 灾难性的; 完全失败的 [?]
  221. discern v. 觉察出; 识别; 了解; 分辨出   [?]
  222. discontinued adj. 停产的 [?]
  223. discontinuous adj. 不连续的 [?]
  224. discrepancy n. 差异; 不符合; 不一致 [?]
  225. discrepancy n. 差异,不一致 [?]
  226. discrepancy n. 差异,不一致,不符合 [?]
  227. discrete adj. 离散的 [?]
  228. discretization n. 离散化 [?]
  229. discretize v. 离散化 [?]
  230. discrimination n. 辨别力,识别力 [?]
  231. discriminative adj. 判别的 [?]
  232. disentangle v. 分清,清理出(混乱的论据、想法等),解开…的结; 理顺 [?]
  233. disparity n. 不同,差异,视差(parallax) [?]
  234. dissertation n. 专题论文; 学位论文 [?]
  235. distill v. 蒸馏,提取;滴下 [?]
  236. distinction n. 差别; 区别 [?]
  237. distinctive adj. 独特的,特别的 [?]
  238. distraction n. 分散注意力的事; 使人分心的事; 娱乐; 消遣 [?]
  239. distress n. 缺陷,瑕疵,困难 [?]
  240. disturbance n. 骚动,动乱 [?]
  241. diverge v. 偏离,分叉,发散 [?]
  242. divergence n. 发散,(意见、态度等的) 分歧,差异 [?]
  243. divergent adj. 有分歧的; 不同的; 相异的 [?]
  244. dodge v. 闪开,躲开,避开 [?]
  245. dominant adj. 首要的; 占支配地位的; 占优势的; 显著的 [?]
  246. dominate v. 支配,控制 [?]
  247. downside n. 缺点,不利方面 [?]
  248. dozens n. 若干 [?]
  249. dramatically adv. 显著地 [?]
  250. drastic adj. 极端的; 急剧的; 严厉的; 猛烈的 [?]
  251. drone n. 无人机 [?]
  252. dummy adj. 假人的 [?]
  253. earnest adj. 非常认真的; 诚实的; 真诚的 [?]
  254. edibility n. 可实用性 [?]
  255. efficacy n. 效力 [?]
  256. egomotion n. 自己的运动 [?]
  257. egomotion n. 自身运动 [?]
  258. elaborate adj. 复杂的,详尽的 [?]
  259. elbow n. 肘 [?]
  260. elegant adj. 优美的; 文雅的; 漂亮雅致的; 陈设讲究的; 精美的; 简练的; 简洁的; 简明的 [?]
  261. eliminate v. 排除,除去 [?]
  262. eliminate v. 消除,消灭 [?]
  263. elliptic adj. 椭圆的 [?]
  264. embellish v. 美化,文章润色 [?]
  265. embodiment n. (体现一种思想或品质的)典型,化身 [?]
  266. emergence n. 出现; 兴起 [?]
  267. emit v. 发射 [?]
  268. empower v. 授权; 给(某人)…的权力; 增加(某人的)自主权,赋能 [?]
  269. emulate v. 仿真 [?]
  270. encapsulate v. 简述; 概括; 压缩 [?]
  271. enclosed adj. 包围的,围住的,封闭的 [?]
  272. encompass v. 包含,包括,涉及(大量事物); 包围; 围绕; 围住 [?]
  273. encompass v. 包括;环绕;围绕 [?]
  274. endeavor n. 努力 [?]
  275. endeavour v. 努力(to do) [?]
  276. enforce v. 强制,坚持 [?]
  277. engineered adj. 设计制造的 [?]
  278. enqueue v. 排队,入列 [?]
  279. enumeration n. 枚举,列举 [?]
  280. environs n. 周围地区 [?]
  281. envision v. 展望; 想象 [?]
  282. ephemeral adj. 短暂的; 瞬息的 [?]
  283. epidemiology n. 流行病学 [?]
  284. equivalence n. 等价,对等 [?]
  285. erroneous adj. 有错误的,错误的 [?]
  286. erroneous adj. 错误的 [?]
  287. erroneously adv. 错误地,不正确地 [?]
  288. evaluation metric  评价指标,评估尺度 [?]
  289. exceed v. 超过 [?]
  290. exceed v. 超过; 超越 [?]
  291. excite v. 激活,使激动; 刺激,触发;引发,引起 [?]
  292. exclusively adv. 仅仅,专门地 [?]
  293. exemplar n. 模范; 榜样; 典型; 范例 [?]
  294. exemplar n. 模范; 榜样; 典型; 范例   [?]
  295. exhaustive adj. 详尽的; 彻底的; 全面的 [?]
  296. exhaustive adj. 详尽的,全面的 [?]
  297. exhaustively adv. 耗尽一切地; 彻底地; 用尽一切地 [?]
  298. exotic adj. 来自异国(尤指热带国家)的; 奇异的 [?]
  299. expertise n. 专长;专门知识;专门技能 [?]
  300. explainability n. 可解释性 [?]
  301. explicit adj. 清楚的,明白的 [?]
  302. explicitly adv. 明确地,清楚地,直接地 [?]
  303. explicitly adv. 明确地;明白地 [?]
  304. explicitly adv. 明确;显然;清楚地;直接地 [?]
  305. exploratory adj. 探索的; 探究的; 探测的; 勘探的   [?]
  306. exponentially adv. 以指数方式 [?]
  307. exteroceptive adj. 外感受性的 [?]
  308. extrapolation n. 外推;外插 [?]
  309. eyebrow n. 眉毛 [?]
  310. facade n. (建筑物的)正面,立面 [?]
  311. facets n. (事物的) 部分,方面; (宝石的) 小平面,琢面   [?]
  312. facilitate v. 促进,促使 [?]
  313. factually adv. 事实上; 确实; 真实; 真实地 [?]
  314. faint adj. (光、声、味) 微弱的,不清楚的; 微小的 [?]
  315. faint adj. 微弱的,不清楚的 [?]
  316. fallacy n. 谬见; 谬论; 谬误 [?]
  317. fashion n. 样式,形式 [?]
  318. fault tolerant adj. 容错的 [?]
  319. feasible  可行的,行得通的 [?]
  320. fidelity n. 忠诚;忠实 [?]
  321. fine-grained adj. 细粒的;有细密纹理的 [?]
  322. flaring n. 火焰,火光 [?]
  323. fleet n. 舰队; 捕鱼船队; (一国的) 全部军舰,海军 [?]
  324. flock n. (羊或鸟) 群; (尤指同类人的) 一大群 [?]
  325. fluctuate v. (大小、数量、质量等) 波动; (在…之间) 起伏不定   [?]
  326. fluctuate v. (大小、数量、质量等)波动; (在…之间)起伏不定   [?]
  327. fluorescent adj. 开花的,花瓣状的 [?]
  328. foothold n. 立足点;立足处(攀登时足可踩的缝、洞、树枝等) [?]
  329. forego v. 先行,在……前面,发生在……之前;放弃,摒弃;拒绝 [?]
  330. formulate v. 用公式表示 [?]
  331. fraction n. 分数,小数部分 [?]
  332. fraction n. 碎片,分数 [?]
  333. geodesy n. 大地测量学,测地学,大地测量 [?]
  334. geography n. 地理,地理学 [?]
  335. gesticulate v. 打手势 [?]
  336. gist n. 要点; 主旨; 大意 [?]
  337. government-subsidized adj. 政府补贴的 [?]
  338. grasp v. 抓取,抓住 [?]
  339. gross adj. 总的,毛的 [?]
  340. grossly adv. 极度地; 极其; 非常 [?]
  341. groundbreaking adj. 开创性的; 创新的; 革新的 [?]
  342. gyroscope n. 陀螺仪 [?]
  343. halved adj. 减半的 [?]
  344. harness v. 利用,控制 [?]
  345. harsh adj. 残酷的; 严酷的; 严厉的; 恶劣的; 艰苦的; 强烈刺眼的; 丑陋的 [?]
  346. heterogeneity n. 异质性;不均匀性;不纯一性;多相性,多样性 [?]
  347. heterogeneous adj. 各种各样的,由很多种类组成的 [?]
  348. heterogeneous adj. 由很多种类组成的; 各种各样的 [?]
  349. heterogeneous adj. 由很多种类组成的; 各种各样的   [?]
  350. heterogeneous adj. 由很多种类组成的; 各种各样的   [?]
  351. heteroscedastic n. 异方差性;异质性;不等变异 [?]
  352. heuristic adj. (教学或教育) 启发式的 [?]
  353. heuristic adj. 启发式的 [?]
  354. heuristics n. 启发式; 探索法 [?]
  355. hierarchical adj. 按等级划分的; 等级制度的 [?]
  356. hierarchical adj. 按等级划分的; 等级制度的 [?]
  357. hierarchy n. 等级制度; 层次体系 [?]
  358. hinder v. 阻碍,阻止 [?]
  359. hip n. 髋 [?]
  360. historical adj. (有关) 历史的; 有关历史研究的; 历史学的 [?]
  361. holistic adj. 全局的,整体的; 全面的 [?]
  362. holistic adj. 整体的; 全面的; 功能整体性的 [?]
  363. hyperbolic adj. 双曲线的 [?]
  364. hyperparameter n. 超参数(人工定义的网络参数,不是通过学习得出,如网络层数、卷积核大小等) [?]
  365. hypothesis n. (有少量事实依据但未被证实的)假说,假设,猜想,猜测 [?]
  366. hypothetical adj. 假设的; 假定的 [?]
  367. imformative adj. 提供有用信息的 [?]
  368. immediate adj. 立即的,立刻的 [?]
  369. immense adj. 巨大的,极大的 [?]
  370. immunization n. 免疫,免疫作用 [?]
  371. impaired adj. 受损的; 损坏的; 出毛病的; 有(身体或智力)缺陷的; 有…障碍的 [?]
  372. imperfection n. 瑕疵,缺点 [?]
  373. implication n. 可能的影响(作用)、牵连 [?]
  374. implicitly adv. 暗中地;不明显地;含蓄地 [?]
  375. imply v. 暗示,包含 [?]
  376. impose v. 推行; 把…强加于; 导致 [?]
  377. impose v. 推行,迫使; 把…强加于 [?]
  378. in-depth adj. 彻底的; 深入详尽的 [?]
  379. inadequacy n. 不足,不够 [?]
  380. incapacitate v. 使失去正常生活(或工作)能力 [?]
  381. incoherent adj. 不清楚的; 难以分辨的; 难懂的; 无逻辑的; 不连贯的 [?]
  382. incompatibility n. 不兼容 [?]
  383. inconclusive adj. 无定论的,不确定的 [?]
  384. inconsistency n. 不一致,矛盾 [?]
  385. incur v. 招致; 遭受; 引起; 引致,带来(成本、花费等) [?]
  386. incure v. 招致,招受 [?]
  387. indicator n. 指示信号; 标志   [?]
  388. inevitable adj. 不可避免的 [?]
  389. inevitable adj. 不可避免的 [?]
  390. inference n. 推断,推理 [?]
  391. inferior adj. 较差的; 次的; 比不上…的; 级别低的; 较低的 [?]
  392. infinite adj. 极大的,无限的 [?]
  393. informative adj. 提供有用信息的; 给予知识的 [?]
  394. inherent adj. 固有的,内在的 [?]
  395. inhibit v. 抑制; 阻止; 阻碍; 使拘束; 使尴尬 [?]
  396. inhomogeneity n. 不均(匀)性;多相(性);同类(质,种)   [?]
  397. inpaint v. 修补,填充 [?]
  398. inset n. 示意图,插图 [?]
  399. insight n. 见识,洞见,洞察,观察 [?]
  400. instantaneous adj. 瞬间的 [?]
  401. instantaneous adj. 立即的,立刻的 [?]
  402. intact adj. 完整的,完好无损的 [?]
  403. integral adj. 必需的; 不可或缺的; 作为组成部分的   [?]
  404. intellectual adj. 智力的; 脑力的; 理智的; 有才智的; 智力发达的 [?]
  405. intension n. 增强;加强;加剧; 内涵;内包; 强度;烈度;紧张 [?]
  406. intensity n. 亮度 [?]
  407. interchangeably adv. 可交替地;可互换地 [?]
  408. interleave v. (尤指将片状物) 插入,夹进 [?]
  409. interpretability n. 可解释性,可解读性 [?]
  410. intractable adj. 很难对付(或处理)的 [?]
  411. intricately adv. 复杂地; 杂乱地 [?]
  412. intriguing adj. 非常有趣的; 引人入胜的; 神秘的 [?]
  413. intuitively adv. 直观地 [?]
  414. intuitively adv. 直观地,直接地 [?]
  415. intuitively adv. 直觉地,直观地;由直觉而得地   [?]
  416. involved adj. 有关的,复杂的 [?]
  417. ionosphere n. 电离层 [?]
  418. irrelevant adj. 无关紧要的; 不相关的 [?]
  419. irritation n. 愤怒 [?]
  420. isoline n. 等值线   [?]
  421. isotropic adj. 各向同性的,等方性的 [?]
  422. jointly adv. 连带地 [?]
  423. kinematics n. 运动学 [?]
  424. laborious adj. 耗时费力的,辛苦的 [?]
  425. latent adj. 潜在的; 潜伏的; 隐藏的 [?]
  426. leap n. 骤变; 剧增; 激增 [?]
  427. lemma n. 引理 [?]
  428. lengthier adj. 很长的; 漫长的; 冗长的 [?]
  429. leverage v. 依靠,使用 [?]
  430. leverage v. 对...施加影响 [?]
  431. limb n. 肢,四肢 [?]
  432. malfunction n. 运转失常; 失灵; 出现故障 [?]
  433. malicious adj. 怀有恶意的; 恶毒的 [?]
  434. maneuver n. 演习; (使某物) 移动,运动; 熟练而巧妙地引导 [?]
  435. maturity n. 成熟 [?]
  436. meander v. 蜿蜒而行; 迂回曲折; 漫步; 闲逛; (乏味地、令人费解地) 漫谈,闲聊 [?]
  437. medicinal adj. 有疗效的 [?]
  438. methodology n. 方法,研究方法 [?]
  439. metric n. 度量,指标 [?]
  440. metropolitan adj. 大城市的; 大都会的; 本土的 [?]
  441. milestone n. 里程碑;转折点;重要事件;重要阶段 [?]
  442. miniature adj. 很小的; 微型的; 小型的 [?]
  443. miniaturization n. 微型化,小型化 [?]
  444. minimalistic adj. 极简化的 [?]
  445. miscellaneous adj. 混杂的; 各种各样的   [?]
  446. miserably adv. 痛苦地;艰苦地;令人难受地; 悲惨地;可悲地; 糟糕地;少得可怜地 [?]
  447. misinterprete v. 误解,误释 [?]
  448. modality n. 形态,形式 [?]
  449. moderate adj. 适度的; 中等的; 温和的; 不激烈的; 不偏激的; 适中的; 合理的   [?]
  450. modestly adv. 谨慎地; 切勿浪费 [?]
  451. modularity n. 积木性;模件性 [?]
  452. monochrome n. 黑白照片,灰度图像 [?]
  453. monotonically adv. 单调地 [?]
  454. mosaicking n. 图像镶嵌,图像拼接   [?]
  455. multiplexing n. 倍增;标度放大 [?]
  456. namely adv. 即; 也就是 [?]
  457. namesake n. 同名的人(或物) [?]
  458. nature n. 特性,特点 [?]
  459. necessitate v. 使...成为必须,使需要 [?]
  460. negligible adj. 微不足道的; 不重要的; 不值一提的   [?]
  461. nocturnal adj. 夜间活动的; 夜间发生的 [?]
  462. noise-free adj. 无噪声的 [?]
  463. nominally adv. 名义上地 [?]
  464. non-deterministic adj. 非确定的; 不确定的 [?]
  465. nonconvex adj. 非凸的 [?]
  466. notably adv. 尤其; 特别 [?]
  467. novel adj. 新颖的,与众不同的 [?]
  468. novelty n. 新颖,创新 [?]
  469. numerous adj. 许多;数量庞大的;数不清的 [?]
  470. nutshell adj. 简洁的;扼要的 [?]
  471. obsolete adj. 淘汰的; 废弃的; 过时的 [?]
  472. obstruction n. 阻挡; 阻碍; 妨碍; 堵塞,阻挡(通道等); 路障; 障碍; 障碍物 [?]
  473. occluded by...  被...阻塞 [?]
  474. occlusion n. 遮挡 [?]
  475. occlusion n. 阻塞,吸留 [?]
  476. off-the-shelf  现货供应的 [?]
  477. omit v. 忽略 [?]
  478. opt v. 选择; 挑选 [?]
  479. optoisolatd adj. 光电耦合的 [?]
  480. ordinal adj. 序数的 [?]
  481. orthogonal adj. 正交的 [?]
  482. oscillating adj. 摇摆,变化; 摆动; 振动; 波动; 振荡 [?]
  483. outperform v. 胜过,比...好 [?]
  484. outperform v. 超过,胜过 [?]
  485. overdetermined adj. 超定的 [?]
  486. overhead n. 天花板,上限,顶端 [?]
  487. overpass v. 过境 [?]
  488. overpower v. 征服,制胜; 令人折服 [?]
  489. overwhelming adj. 巨大的;压倒性的;无法抗拒的 [?]
  490. palette n. 调色板   [?]
  491. panchromeatic adj. 全色的 [?]
  492. pandemic adj./n. (疾病)流行全国[全世界]的/(流行全国[全世界]的)传染病,瘟疫 [?]
  493. parabolic adj. 抛物线的,抛物线状的 [?]
  494. paradigm n. 典范; 范例; 样式 [?]
  495. paradigm n. 模型(正式) [?]
  496. parallax n. 视差 [?]
  497. parallel adj. 平行的; 极相似的 [?]
  498. parallelism n. 相似; 相似的特点 [?]
  499. parameterization n. 参数化 [?]
  500. pathway n. 途径; 路径; 小路; 行动路线 [?]
  501. patrol n. 巡逻 [?]
  502. paucity n. 少量; 少许; 贫乏 [?]
  503. payload n. 装载量,负载 [?]
  504. pedestrian n. 行人,路人 [?]
  505. penalize v. 处罚; 惩罚 [?]
  506. penalize v. 惩罚,处罚 [?]
  507. penalty n. 惩罚 [?]
  508. pendulum n. 钟摆 [?]
  509. percentile n. 百分数 [?]
  510. perceptron n. 感知器 [?]
  511. perennial adj. 长久的; 持续的; 反复出现的; 多年生的 [?]
  512. periodic adj. 周期的,间发性的,定期的 [?]
  513. periodic adj. 间发性的; 定期的; 周期的 [?]
  514. persistent n. 执著的; 不屈不挠的; 坚持不懈的; 连绵的; 持续的; 反复出现的 [?]
  515. persuasively adv. 有说服力地; 令人信服的 [?]
  516. perturbation n.  摄动; 微扰; 小变异,扰动 [?]
  517. perturbation n. 摄动;扰动;微扰;忧虑 [?]
  518. perturbed adj. 被扰乱的 [?]
  519. pervasive adj. 遍布的; 充斥各处的; 弥漫的 [?]
  520. photoelectron n. 光电子 [?]
  521. piecemeal adj. 零散的 [?]
  522. pillar n. 柱子 [?]
  523. pilot v. 驾驶 [?]
  524. pipeline n. 流程,管线 [?]
  525. pitfall n. 危险; 困难; (尤指)陷阱,隐患   [?]
  526. pivotal adj. 关键性的; 核心的 [?]
  527. planimetric adj. 水平投影的 [?]
  528. plausible adj. 似是而非的;似乎可信的;似乎合理的 [?]
  529. plethora n. 过多,过剩 [?]
  530. poisoning n. 中毒 [?]
  531. pole n. 杆,电线杆 [?]
  532. polynomial-time  多项式时间 [?]
  533. posterior adj. 后验的 [?]
  534. postpon v. 延期,推迟 [?]
  535. practitioner n. 从业人员 [?]
  536. precaution n. 预防;预防措施 [?]
  537. preceding adj. 先前的,前面的 [?]
  538. preclude n. 前面的步骤,前奏;序幕 [?]
  539. preintegrated adj. 预先整合的 [?]
  540. preparedness n. 准备好的状态;愿意 [?]
  541. prerequisite n. 前提,先决条件 [?]
  542. presence n. 存在,出现 [?]
  543. presence n. 存在,出现 [?]
  544. presence n. 存在,在场,出席 [?]
  545. presumably adv. 很可能; 大概; 想必是 [?]
  546. prevalence n. 流行 [?]
  547. prey v. 捕食 [?]
  548. priori adj. 先验的 [?]
  549. prioritize v. 按重要性排列; 划分优先顺序; 优先处理   [?]
  550. pro n. 赞成的意见,优点 [?]
  551. problematic adj. 造成困难的; 产生问题的 [?]
  552. problematic adj. 造成困难的; 产生问题的   [?]
  553. procurement n. (尤指为政府或机构)采购 [?]
  554. prohibitely expensive adj. 价格高得离谱的 [?]
  555. prominent adj. 重要的; 显眼的; 显著的; 突出的; 凸现的 [?]
  556. prominent adj. 重要的; 著名的; 杰出的 [?]
  557. promising adj. 有前景的;大有希望的 [?]
  558. promising adj. 有希望的,有前途的 [?]
  559. prone adj. 易于遭受…的 [?]
  560. propagate v. 传播,宣传 [?]
  561. propeller n. 推进器 [?]
  562. proportional adj. 成比例的 [?]
  563. proposition n. 命题 [?]
  564. proposition n. 提议,建议,待处理的问题 [?]
  565. proprietary adj. 专卖的; 专营的; 专利的; 所有的; 所有权的 [?]
  566. proprietary adj. 专卖的,专利的 [?]
  567. proprioceptive adj. 本体感觉; 本体感受的 [?]
  568. propulsion n. 推进;推动力 [?]
  569. protocol n. 议定书;始末记;(条约等的)草案;协议 [?]
  570. proximity n. (时间或空间) 接近,邻近,靠近 [?]
  571. proxy n. 代理,代理人,(测算用的) 代替物,指标 [?]
  572. punchy adj. 简洁有力的; 言简意赅的; 简练的 [?]
  573. purview n. (组织、活动等的) 范围,权限 [?]
  574. putative adj. 推定的; 认定的; 公认的 [?]
  575. quadratic adj. 平方的; 二次方的 [?]
  576. quadrilateral n. 四边形 [?]
  577. qualitative adj. 质量的,定性的 [?]
  578. quantitative adj. 定量的 [?]
  579. quantize v. 量化 [?]
  580. quasi adj. 准的,类似的,外表的   [?]
  581. reaction whell n. 反作用轮 [?]
  582. realistic adj. 现实的,实际的 [?]
  583. reckoning n. 估计; 估算; 计算 [?]
  584. reconnaissance n. 侦查 [?]
  585. recurrent adj. 循环的,反复出现的 [?]
  586. recursive adj. 递归的,循环的 [?]
  587. recursive adj. 递归的,循环的 [?]
  588. redundancy n. 冗余;冗余度 [?]
  589. regime n. 组织方法; 管理体制 [?]
  590. regularization n.  正则化,规范化 [?]
  591. regulatory adj. 管理的,监管的 [?]
  592. regulatory adj. 管理的;调整的;监管的 [?]
  593. relaxation n. 简化,松弛 [?]
  594. relay v. 转播;传达 [?]
  595. relevancy n. 相关性; 关联性; 关系; 关联 [?]
  596. remit n. 职权范围; 控制范围; 影响范围 [?]
  597. renewed adj. 再次发生的; 再次兴起的; 更新的 [?]
  598. reproducibility n. 重复性,复现性 [?]
  599. requisite n. 必须的事物,先决条件 [?]
  600. resemble v. 像,看起来像 [?]
  601. resilience n. 快速恢复的能力; 适应力; 还原能力; 弹力 [?]
  602. restrictive adj. 限制性的; 约束的 [?]
  603. resurgence n. 复苏; 复兴 [?]
  604. resurgence n. 复苏; 复兴 [?]
  605. retain v. 保留,持有 [?]
  606. retention n. 保留,保持 [?]
  607. retrieval n. 检索 [?]
  608. retrieval n. 检索 [?]
  609. rig n. 装配,配置 [?]
  610. ripple n. 波浪 [?]
  611. road grader n. 压路机 [?]
  612. rotary adj. 旋转的; 绕轴转动的; 转动的 [?]
  613. sacrifice n. 牺牲 [?]
  614. sailboat n. 帆船 [?]
  615. saliency n. 显著性 [?]
  616. saliency n. 显著,特点,显著性 [?]
  617. salient adj. 显著的,突出的 [?]
  618. salient adj. 显著的,突出的 [?]
  619. salient adj. 最重要的; 显著的; 突出的 [?]
  620. salient adj. 最重要的;显著的;突出的 [?]
  621. saturated adj. 饱和的,饱满的 [?]
  622. saturation n. 饱和;饱和度;饱和状态 [?]
  623. scarcity n. 缺乏; 不足; 缺少 [?]
  624. scenario n. 场景 [?]
  625. schematic adj. 略图的; 简表的 [?]
  626. schematic n. 简图,原理图 [?]
  627. seamless adj. 无缝的 [?]
  628. sector n. 行业 [?]
  629. self-contained adj. 独立的,自主的 [?]
  630. semidefinite adj. 半正定的 [?]
  631. semiotics n. 符号学 [?]
  632. sensor-agnostic adj. 传感器未知的 [?]
  633. sensory adj. 感觉的,感官的 [?]
  634. servo-mounted adj. 伺服安装的 [?]
  635. session n. 一场; 一节; 一段时间 [?]
  636. severity n. 严重性;严重度;严重程度 [?]
  637. shorthand n. 速记(法);(对某事)简略的表达方式 [?]
  638. shrinkage n. 收缩;缩小 [?]
  639. shrinkage n. 缩小; 收缩 [?]
  640. siamese adj. 孪生的 [?]
  641. siamese adj. 孪生的,相似的 [?]
  642. silent adj. 安静的,寂静的 [?]
  643. silhouette n. 人的剪影、轮廓 [?]
  644. sinusoidal adj. 正弦曲线的 [?]
  645. skeletal adj. 骨架的,骨骼的 [?]
  646. skewed adj. 歪曲的; 不准确的 [?]
  647. skewed adj. 歪曲的; 有偏颇的; 不准确的; 偏向(或偏重)…的 [?]
  648. skyrocket v. 飞涨,猛涨 [?]
  649. sluggishly adv. 慢腾腾; 缓慢地; 行动迟缓地; 迟缓地; 慢吞吞地 [?]
  650. snippet n. 一小条(消息); 一则(新闻); 一小段(谈话、音乐等) [?]
  651. sobering adj. 令人警醒的; 使人冷静的 [?]
  652. socioeconomic adj. 社会经济的 [?]
  653. soldering n. 焊锡 [?]
  654. sole adj. 仅有的,唯一的,独有的 [?]
  655. solely adv. 仅仅; 只; 唯; 单独地 [?]
  656. solidarity n. 团结;齐心协力;同心同德;相互支持 [?]
  657. sparsify v. 稀疏化 [?]
  658. specification n. 规范,规格,说明书 [?]
  659. specularity n. 镜面反射; 高光 [?]
  660. speedboat n. 快艇 [?]
  661. spider v. 爬虫 [?]
  662. spiky adj. 有尖刺的; 刺猬式的 [?]
  663. spoof v. 恶搞;哄骗;诳骗 [?]
  664. spuriou adj. 虚假的,伪造的 [?]
  665. spurious adj. 虚假的; 伪造的 [?]
  666. stain n. 污点; 污渍; 染色剂; 着色剂; (名声上的) 污点   [?]
  667. state v. 表达,陈述 [?]
  668. state-of-the-art  最新的 [?]
  669. status quo  现状,维持现状 [?]
  670. steadfast adj. 坚定的;不动摇的 [?]
  671. stimuli n. 刺激物; 促进因素; 激励因素; (使生物产生反应的)刺激 [?]
  672. stochastic adj. 随机的,随便的,有可能性的 [?]
  673. stochastic adj. 随机的; 机会的; 有可能性的; 随便的   [?]
  674. strategic adj. 根据全局而安排的;战略性的;战略上的 [?]
  675. stratified adj. 社会分层的; 等级分化的 [?]
  676. stray adj. 零星的; 孤立的; 离群的; 走散的 [?]
  677. striking adj. 引人注目的; 异乎寻常的; 显著的 [?]
  678. strikingly adv. 极; 醒目地 [?]
  679. subdivide v. 再分,细分 [?]
  680. suboptimal adj. 次优的,次佳的 [?]
  681. subordinate adj. 从属的; 下级的; 次要的 [?]
  682. substantial adj. 大量的,巨大的 [?]
  683. substantially adv. 非常; 大大地 [?]
  684. substitute into  代入... [?]
  685. subtle adj. 微妙的,巧妙的; 不易察觉的,不明显的 [?]
  686. subtotal n. 小计;部分和 [?]
  687. succession n. 一连串; 一系列; 连续的人(或事物)   [?]
  688. successive adj. 连续的; 接连的; 相继的 [?]
  689. successor n. 继任者,接替者 [?]
  690. suffice v. 足够 [?]
  691. suffice v. 足够; 足以 [?]
  692. suitability n. 适合 [?]
  693. superimpose v. 使(图像甲)叠映在(图像乙)上; 使重叠; 使叠加; 使附加于 [?]
  694. superposition n. 叠加态; 叠加性; 疊加; 迭加; 叠加法 [?]
  695. supervision n. 监督,管理 [?]
  696. surrogate n. 代替 [?]
  697. surrogate n. 替代品,代替 [?]
  698. susceptible adj. 易受影响(或伤害等); 敏感; 过敏; 好动感情的; 感情丰富的 [?]
  699. swath n. 卫星的一轨,轨道 [?]
  700. swaying adj. 摇摆的 [?]
  701. swiftlet n. 金丝燕 [?]
  702. synchronize v. 同步 [?]
  703. synonyms n. 同义词 [?]
  704. syntax n. 语法,句法,句法规则   [?]
  705. synthesize v. 人工合成 [?]
  706. synthesize v. 合成 [?]
  707. synthesized adj. 合成的 [?]
  708. synthetic adj. 人造的,人工的 [?]
  709. synthetic adj. 人造的,合成的 [?]
  710. synthetic adj. 合成的 [?]
  711. synthetically adv. 人工合成地 [?]
  712. sythesized adj. 合成的 [?]
  713. tabular adj. 表格的,列成表的 [?]
  714. tackle v. 处理 [?]
  715. tactile adj. 触觉的; 有触觉的; 能触知的 [?]
  716. taxonomy n. 分类学; 分类法; 分类系统 [?]
  717. tedious adj. 冗长的 [?]
  718. tedious adj. 冗长的; 啰唆的; 单调乏味的; 令人厌烦的 [?]
  719. telemetry n. 遥测,测控 [?]
  720. temporal adj. 时间的 [?]
  721. temporally adv. 时间上的 [?]
  722. temporally synchronized  时间同步的 [?]
  723. tendency n. 倾向; 偏好; 性情; 趋势; 趋向 [?]
  724. terminology n. (某学科的)术语; 专门用语 [?]
  725. thereafter adv. 之后,此后,以后 [?]
  726. thereof adv. 在其中; 由此 [?]
  727. thesaurus n. 分类词典,同义词词典 [?]
  728. thesis n. 论文(学位论文),区别于paper [?]
  729. thorough adj. 彻底的 [?]
  730. thoroughfare n. 大街; 大道 [?]
  731. threefold adj. <正>三倍(的); 有三部分的,三重的 [?]
  732. thrust v. 刺,塞 [?]
  733. tilde n. 波浪号 [?]
  734. tint n. 色调 [?]
  735. toilsome adj. 辛苦的 [?]
  736. topographic adj. 地形的 [?]
  737. trace n. 痕迹; 遗迹; 踪迹 [?]
  738. tractable adj. 易处理的; 易驾驭的 [?]
  739. tractable adj. 易处理的,易驾驭的 [?]
  740. tractable adj. 易处理的;易驾驭的 [?]
  741. tradeoff n. 权衡,折中,取舍 [?]
  742. trajectory n. 轨迹,弹道 [?]
  743. transduction n. 转导; 换能,转换,变频 [?]
  744. transect n. 横切; 断面; 横断面   [?]
  745. transient adj. 短暂的;转瞬即逝的 [?]
  746. transition n. 过渡; 转变; 变革; 变迁 [?]
  747. transition n. 过渡; 转变; 变革; 变迁 [?]
  748. trigonometric adj. 三角函数的 [?]
  749. trilateration n. 三边测量; 三边测量法; 三边测量定位; 三边定位; 定位 [?]
  750. trivial adj. 不重要的,琐碎的 [?]
  751. trivial adj. 不重要的,琐碎的 [?]
  752. trivial adj. 琐碎的; 微不足道的; 不重要的,无关紧要的 [?]
  753. trivially adv. 十分 [?]
  754. tubular adj. 管道的,管状的 [?]
  755. ubiquitous adj. 似乎无所不在的; 十分普遍的 [?]
  756. ubiquitous adj. 十分普遍的 [?]
  757. ubiquitous adj. 无处不在的,十分普遍的 [?]
  758. ultimately adv. 最后地,根本上地 [?]
  759. underdeterminted adj. 欠定的 [?]
  760. underestimate v. 低估; 对…估计不足; 对…认识不足(或重视不够); 轻视 [?]
  761. underneath prep. 在…底下,下面 [?]
  762. underscore v. 强调; 着重说明 [?]
  763. unleash v. 发泄; 突然释放; 使爆发 [?]
  764. unprecedented adj. 前所未有的 [?]
  765. upheaval n. 剧变; 激变; 动乱; 动荡 [?]
  766. utilization n. 使用,利用 [?]
  767. utilization n. 利用,应用 [?]
  768. utilize v. 使用,利用 [?]
  769. vacuum v. 用吸尘器清扫 [?]
  770. vague adj. 含糊的,不明确的 [?]
  771. vague adj. (思想上) 不清楚的,含糊的,不明确的,模糊的; 不具体的; 不详细的   [?]
  772. vaguely adv. 模糊地 [?]
  773. valuable adj. 有用的,重要的 [?]
  774. vendor n. 小贩; 摊贩; (某种产品的) 销售公司; (房屋等的) 卖主 [?]
  775. verbatim adv. 一字不差的(地); 逐字的(地) [?]
  776. versatile adj. 多才多艺的;有多种技能的;多面手的;多用途的 [?]
  777. viable adj. 切实可行的; 可实施的; 能独立发展的; 能独立生存的; 可生长发育的; [?]
  778. viable adj. 可实施的; 切实可行的; 能独立发展的; 能独立生存的; 可生长发育的 [?]
  779. vibrant adj. 充满生机的 [?]
  780. vignetting n. 暗角; 渐晕; 虚光; 晕影; 黑角 [?]
  781. violated adj. 污损的 [?]
  782. vulnerable adj. 脆弱的,易受...影响的 [?]
  783. war-torn adj. 遭受战争的,被战争损坏的 [?]
  784. warehouse n. 仓库 [?]
  785. wire n. 线,电线 [?]
  786. wrist n. 手腕 [?]
  787. yet conj. 但是; 然而 [?]
  788. flurry n. 一阵 [?]
↑ return to top ↑

Expressions

  1. , which results in delays in processing 导致了在处理上的延迟/滞后 
  2. ... achieve very high resolution imagery at low cost 以低成本实现了很高地表分辨率的影像
  3. ... achieves better performance than all the end-to-end methods by a notable margin. ...相比于其它端到端方法的效果好了一大截(有显著的提升)。
  4. ... are available for every month from April 2012 onward. 从2012年4月往前,每个月都可获取
  5. ... are detailed in part 3. ...在第三部分进行了详细描述
  6. ... are highly distinctive and robust to changes in viewing conditions ...具有很高的区分性,并且对场景变化很鲁棒
  7. ... are inadequate for ... ...对...是不足(够)的
  8. ... are mainly caused by ... ...主要由...引起
  9. ... are not limited to ...不局限于 
  10. ... are parallel to ... ...与...相似的(平行的)
  11. ... are screened out by thresholding ...可以通过阈值筛选而去除
  12. ... are summarized in Table 1 ...总结如表1
  13. ... are summarized next ...总结如下
  14. ... at the expense of a slightly higher computational cost ...伴随着一点点更高的计算开销
  15. ... be classified into ... ...
  16. ... bridge the ... gap between ... and ... ...架起了...和...之间的...的桥梁
  17. ... but still requires lots of research works ...但仍需要大量的研究工作
  18. ... by a large margin 一大截
  19. ... calculated by concatenating all the predicted frame-to-frame poses between keyframe 通过将所有预测的关键帧到关键帧的位姿连接在一起来计算...
  20. ... can be represented either explicitly or implicityly ...可以被显式或隐式地表达
  21. ... can detect radiances down to 1.54 and up to 3.17 ...可以检测下到1.54上到3.17的辐射
  22. ... can introduce errors ...可以引入误差
  23. ... can only estimate camera egomotion and environmental structure up to an arbitrary scale factor ...仅仅只能够估计出随机任意尺度的相机的自身运动和环境结构
  24. ... can provide better accuracy than ... and much lower resources utilization than ... 相比于...和...,...可以提供更好的精度和更低的资源利用
  25. ... confirms the superiority of our method ...确认了我们方法的优越性
  26. ... contains most of common daily scenes in which a person usually appear ... 包含了一个人通常出现在日常生活中的大多数场景
  27. ... deliver stable registration results ...得到稳定的配准结果
  28. ... does not give sufficient accuracy ...不能给出足够的精度
  29. ... excessive amounts of motion blur will be captured. 会捕捉到大量的运动模糊
  30. ... extremely complex and expensive 尤其复杂与昂贵
  31. ... fill a gap between in introductory-level robotics courses and extensible drone platforms for education ... 填补入门级机器人技术课程与可扩展无人机教育平台之间的空白
  32. ... gives the most efficient performance, although it makes a significant sacrifice in overall accuracy, as well as robustness on challenging trajectories. ... 提供最有效的性能,尽管它在整体精度上做出了重大牺牲以及牺牲了在具有挑战性的轨迹上的鲁棒性。
  33. ... has a large overlap with ... 与...有很大重叠
  34. ... has attracted significant interest in both the computer vision and robotics communities over the past few decades 在过去十几年中,...在计算机视觉和机器人领域中已经吸引了大量的关注
  35. ... has been widely applied in ... ...已经被广泛应用在...
  36. ... has received a high degree of attention and research efforts over the last years. ... 在过去的几年里受到了高度的关注和研究。
  37. ... has received noticeably less attention than ... ...受到了比...相当少的关注
  38. ... have an extensive collection of topics ...涉及到广泛的话题
  39. ... have been documented on the ground in China 已经在中国实际落地
  40. ... have no overlap with ... ...和...之间没有重叠
  41. ... have received a great deal of attention ...已经引起/吸引了大量的关注
  42. ... have some overlap in the along track direction, ant there is a small across-track overlap ... 在沿轨方向有一些重叠,同时在垂轨方向也有小部分重叠
  43. ... help in finding solutions to close the gap in existing research ...帮助找到缩小目前研究(和实际)差距的方案
  44. ... is a continuation of ... by other means ...是...的另一种形式的延伸
  45. ... is another important aspect of a camera model ...是相机模型另外一个重要的方面
  46. ... is approximately nine times slower than ... ...大约比...慢了9倍
  47. ... is depicted in Figure 7. ...在图7中进行了说明
  48. ... is essentially a state estimation problem ...本质上说是一个状态估计问题
  49. ... is inversely proportional to cell width at that level ...与那一层的格网宽度成反比
  50. ... is most closely related to our work ...与我们的工作最为相关
  51. ... is not optimized for real-time operation ...并没有针对实时运行进行优化
  52. ... is one of the most important metrics for a camera model. ...是相机模型重要的评价指标之一
  53. ... is orders of magnitude higher than frame rate ...比帧率快(高)了几个数量级
  54. ... is particularly beneficial for ... ...对...尤其有益
  55. ... is provided in Figure 9 ...如图9所示
  56. ... is the first commercial microsatellite that can produce sub-meter GSD imagery and video when combined with ground-based image processing systems. ...是第一颗商业微纳卫星,当与地面图像处理系统结合时,可以产生亚米级地表分辨率的图像和视频。
  57. ... is the ideal resolution to detect activity that impacts the global economic landscape ...是用于探测影响全球经济活动的理想的地表分辨率
  58. ... is the process of calculating current location at time t by using a previously determined position at time t − 1 by using estimations of speed and course over elapsed time. 根据过去的速度和航线求解得到t-1时刻的位置,...就是根据之前t-1时刻的位置求解当前t时刻的位置的过程。
  59. ... is used to compensate for ... ...被用来补偿...
  60. ... is vastly superior to ... ...比...优越(好)了很多
  61. ... is with a swath width of ~3000 km and two spectral bands ...幅宽大约3000公里,有两个光谱波段
  62. ... large pixels to collect many photoelectrons 用于手机很多光电子的大像素
  63. ... many are specific to a certain type of hardware or are limited to a particular environment. ...许多都针对于特定类型的硬件或仅限于特定环境
  64. ... marked the start of a new era in NTL data collection and applications. ...标志着NTL数据和应用新的时代的开始
  65. ... may not be persented and utilized properly ...可能没有被很好的展现和利用 
  66. ... may pose a restriction in practice ...可能会给实际应用带来限制
  67. ... poses challenges due to three key issues originating from the satellite 由于卫星方面的三个关键问题,...带来了些许挑战
  68. ... present large errors ...引入了很大的误差
  69. ... relaxes the pointing stability required by ... ...缓解了由...带来的对指向稳定性的要求
  70. ... resolutions grater than 1m GSD 优于1m的地表分辨率
  71. ... run at a very high frame rate 以一个比较高的帧率运行
  72. ... seems to be a trivial detail but turns out rather crucial. ...似乎是一个不重要的细节但事实上却很重要
  73. ... sheds some light for future research ...为未来的研究带来了一些光明
  74. ... solely based on ... ...仅仅基于...
  75. ... squeeze very high-performance into a very small form factor(mass and volume) ... 将非常高的性能压缩到非常小的形状因子(质量和体积)
  76. ... suffer from an inherent lack of scale information ...面对尺度信息的先天缺乏情况
  77. ... surpasses RCF, HED, and FCN in terms of AIU 在AIU指标上而言,...比RCF、HED、FCN表现更好
  78. ... theoretically limit ...
  79. ... uniform pixels responses across all three sensors 在所有三个传感器中统一像素响应
  80. ... use the absolute trajectory RMSE as the error metric for our experiments ...用绝对轨迹RMSE作为实验的误差度量
  81. ... we do not cover the algorithm in depth 我们不深入讨论算法
  82. ... which is the foundation of benchmarking the accuracy of different algorithms ...是衡量不同算法准确性的基础
  83. ... which leaves many open issues ...留下许多悬而未决/待解决的问题
  84. ... will also arise ...也会出现
  85. ... with increasing accuracy and robustness ...伴随着日益增长的精度与鲁棒性
  86. ... without incurring motion blur or smear into the imagery 不会引起图像的运动模糊或模糊
  87. ... work on par with classical methods ...与传统经典方法平分秋色/不分上下
  88. ..., followed by ... ...伴随着...
  89. 45 times higher than OLS 比OLS高了45倍
  90. A fairly common assumption in the current localization approach is that the world remains unchanged as the robot moves through it. 当前定位方法中一个相当常见的假设就是当机器人在某个场景中运动的时候,这个场景的其它部分是保持不变的(静止的)。
  91. A map is necessary to localize, but one must localize to create a map. - chicken and egg problem 地图对于定位来说是必须的,但同时又需要定位才能创建地图-鸡生蛋蛋生鸡的问题。
  92. A real-time requirement meant a slight reduction in accuracy in exchange for constant time assumption. 实时性要求意味着用不变的时间假设来交换精度的轻微降低。
  93. A study of ... was conducted to determine the overall accuracy of WorldView ortho products. 为了确定WorldView正射产品的总体精度,一个...研究被进行了。
  94. Ablation study 烧蚀研究(利用控制变量法看看不同网络层对最终结果造成的影响)
  95. Absolute Trajectory Error(ATE) 绝对轨迹误差
  96. Absolute geolocation accuracy is a measure of the location of an object, as it appears in a product, with respect to its true location on the Earth. 绝对地理定位精度是指对影像中出现的某个物体的位置与其在地球上真实位置的测量。
  97. Accordingly, ... 相应地...
  98. Accuracy and robustness can be improved with additional computation, but on systems with limited resources, finding the right balance between the computing requirements can be challenging. 通过额外的计算可以提高精度和鲁棒性,但对于资源有限的系统,在计算需求之间找到正确的平衡点可能是一项挑战。
  99. Achieving both high temporal resolution and high spatial resolution has been very hard/difficult to achieve. 实现高时间分辨率和高空间分辨率一直是非常困难的。
  100. Additional information must be exploited to resolve the metric scale of the solution. 必须利用额外的信息来解决尺度的不确定性。
  101. Although our method is not real time, the speed can be improved by the development of hardware and the technology of model compression. 虽然我们的方法不是实时的,但是运行速度可以随着硬件和模型压缩技术的发展而不断提高。
  102. Although the assumption holds for some robotic applications. 尽管这种假设适用于一些机器人应用。
  103. Among motion based approaches 在基于运动的方法中...
  104. Arriving at a similar conclusion 取得/得到了相似的结论
  105. As a fundamental task in image processing, image registration aligns two or more images with overlapping scenes taken at different times, from different viewpoints, or by different sensors. 作为影像处理中的基础任务,影像配准将两个或多个不同传感器、不同视角、不同时间拍摄的具有共同区域的影像进行对齐。
  106. As a result, pictorial cues have seen relatively little use in these scenarios until recently. 作为结果,目前图像线索已经被较少使用在这些场景中了。
  107. As expected 和期望一样
  108. As it is, 事实上
  109. As robots become more common, they also become more accessible and easier to develop. 随着机器人越来越普遍,它们也变得更容易获取和开发。
  110. As the camera platform moves, the view plane sweeps out a region of interest. A 2-D image is formed by the ensemble of all captured 1-D images. 随着相机平台的移动,在视平面上同样会扫出一个对应区域。一个2D影像就可以由所有捕获到的1D影像合成。
  111. As the segment contours are high-gradient areas, the keypoints falling in this intersection have to be removed. 由于线段轮廓是高梯度区域,因此必须删除落在这些区域上的特征点。
  112. BRIEF descriptor is a bit string description of an image patch constructed from a set of binary intensity tests. BRIEF描述子就是一个对某一块影像进行描述的二进制字符串,由一系列的灰度测试结果构成。
  113. Because of both practical and theoretical benefits, 由于实际与理论上的好处...
  114. Because of the advantages and disadvantages of both methods, we consider that they are complementary and therefore their combined used is an effective way of achieving accurate tracking and mapping. 由于这两种方法的优缺点,我们认为它们是互补的,因此它们的结合使用是实现精确跟踪和建图的有效途径。
  115. Because the segment contours usually become high-gradient areas, salient point features tend to appear. We do not consider the features in such contour areas. 由于线段轮廓通常成为高梯度区域,因此会出现显著的点特征。我们不考虑这些边缘区域的特征。
  116. Benefiting from the revisit of remote sensing sensors 得益于遥感传感器(对同一区域)的重访
  117. But it requires specialized hardware and increases the complexity of integration. 但是它需要特殊的硬件,并且增加了集成的复杂度。
  118. But sometimes, this assumption may not be the case. 但有时候,这个假设并不一直成立。
  119. But the localization result for GPS chips is bad with meters of error. 但是GPS芯片的定位结果比较差,有几米的误差。
  120. By using a set of local observations, a balance between computational cost and accuracy can be found. 通过很多局部观测,我们可以找到一个计算消耗和精度之间的平衡。
  121. CE90 is the circular error at the 90th percentile. This means that a minimum of 90 percent of the points measured has a horizontal error less than the stated CE90 value. CE90是第90百分位处的圆误差。这意味着至少90%的测量点的平面误差小于规定的CE90值。
  122. CPU and memory load CPU与内存负载
  123. Cameras and IMU are a natural complementary set of sensors. 相机与IMU是天然的互补传感器。
  124. Comparison of poses from our approach against those of Stereo ORB-SLAM2 on two test trajectories. 我们方法与双目ORB-SLAM2在两个测试轨迹上的位姿比较。
  125. Considering the limitation of orthorectification that requires accurate sensor geometry and high-precision DEM, we employ a PL transform to fit the local distortions caused by terrain relief. 考虑这个需要精确传感器参数和高精度DEM的正射校正方法的限制,我们通过使用PL变换来拟合由地形起伏而导致的局部的变形。
  126. Data values are reported in digital numbers (DN) on a sixbit scale that ranges from 0 (no light) to 63 (maximum reported light). 以6bit数值(DN)记录的数据值数值范围在0到63之间。
  127. Depending on the accuracies of ... 依赖于...的精度
  128. Designing an algorithm that is robust under a wide variety of scenes encountered in complex real-life applications remains an open problem. 设计一个在复杂的实际生活中可以遇见的、不同场景下的鲁棒的算法仍然是一个悬而未决的问题。
  129. Despite the simplicity of our method, 尽管我们的方法比较简单
  130. Due to space constraint 由于空间限制
  131. Due to the power and payload constraints of aerial platforms, state estimation algorithms must provide these qualities under the computational constraints of embeded haraware. 由于空中平台的功率和有效载荷的限制,状态估计算法必须在嵌入式硬件的计算限制下提供这些特性。
  132. Each of the channel captures a different aspect of information. 每一个通道都捕获信息的不同的方面。 
  133. Early approaches to tackle this issue ... 解决这个问题的早期方法...
  134. Early filter-based approaches have quickly evolved to sophisticated, hierarchical, factor graph-based optimizations. 早期基于滤波的方法很快地进化到了复杂的、按等级划分的、基于因子图的优化。
  135. Earth-observing sensor 地球观测传感器
  136. Evaluation on a test set shows that it works correctly in those cases, but it does not guarantee correct behavior in other scenarios. 在一个测试集上的评价显示在这些场景中它可以很好工作,但这并不能保证它在其它场景中也可以正确运行。
  137. Even though, ORB-SLAM3 is the best performing system in the outdoor sequences. 即便如此,ORB-SLAM3仍然是户外序列中最好的系统。
  138. Experiments on video data and handwritten digits further demonstrate the generality and broad applicability of our method. 在视频数据和手写数字的实验,进一步说明了我们方法的泛化性和应用的广泛性。
  139. Extensive experiments are conducted to demonstrate the superiority and generalizability of FPHBN. (我们做了)大量的实验证明FPHBN的优越性和通用性。
  140. Extensive experiments on benchmark and additional challenging datasets demonstrate that our method significantly outperforms the state-of-the-art approaches and works effectively on a wide range of complex scenarios. 在基准和额外挑战性数据集上的大量实验显示,我们的方法比目前最先进的方法效果有显著提升,并且在多种复杂场景中都可以高效工作。
  141. FPHBN achieves best mean ODS and OIS and surpasses the second best by a large margin. FPHBN取得了最好的ODS和OIS结果,并且比第二好的结果好了一大截。
  142. FPHBN achieves superior performance compared to HED 相比于HED,FPHBN取得了更好的结果/表现
  143. Figure best viewed in electronic format. 以电子格式查看图片有最好效果。
  144. Figures are best viewed in color. 在彩色模式下看图有最好效果。
  145. Finally, future directions for NTL research are suggested. 最后,介绍了NTL研究的未来方向。
  146. For a single camera, 3D points that are in the same direction but at different distances are projected to the same 2D point, which leads to the scale ambiguity. When a second camera with a constant transformation relative to the first one is added, the scale ambiguity is eliminated. 对于单摄像机,同一方向不同距离的三维点都会投影到同一个二维点上,导致尺度不确定。当添加相对于第一个相机具有恒定变换的第二个相机时,消除了这种尺度不确定。
  147. For a target number N of keypoints 对于N个目标特征点
  148. For all entries 对于所有项目/条目
  149. For an individual control point, the geolocation error is measured by calculating the difference between the observed location in the product and the known, surveyed location. 对于单个控制点,通过计算产品中观测位置与已知测量位置之间的差异来测量地理位置误差。
  150. For better discriminative power 为了更好的分辨能力
  151. For each reference image, we compute the oFAST keypoints and rBRIEF features, targeting 500 keypoints per image. 对于每个参考影像,我们计算oFAST角点和rBRIEF描述子,提取500个特征点。
  152. For maximum computational efficiency 为了获得最高计算效率
  153. For that reason 因为那个原因
  154. Fortunately, 幸运的是
  155. From a theoretical standpoint, 从理论的角度
  156. Furthermore, we have compared our system against the original ORB-SLAM2 to quantify the improvement of our approach in dynamic scenes. 此外,我们还将我们的系统与原始的ORB-SLAM2进行了比较,以量化我们的方法在动态场景中的改进。
  157. GT ground-truth
  158. Geolocation accuracy is determined by comparing a known, surveyed location (typically a ground control point), to the corresponding photo-identifiable feature in an image product.  地理定位精度是通过将已知的测量位置(通常是地面控制点)与图像产品中相应的照片可识别特征进行比较来确定的。
  159. Geolocation accuracy is driven by the sensors and models used in the imagery collection system and by terrain displacement when the image pixels are projected to a surface on the Earth. 地理定位精度由图像采集系统中使用的传感器和模型决定,也受图像像素投影到地球表面时的地形位移影响。
  160. Geolocation accuracy of ortho products is determined by creating ortho products over sites that contain a dense network of surveyed ground control points.  正射产品的地理定位精度是通过在包含已测地面控制点密集网络的场地上创建正射产品来确定的。
  161. Grand total 总计
  162. HS imagery registration is a crucial prelude to many highlevel HS tasks. 高光谱影像配准对于高层次的任务来说是关键的前序步骤。
  163. Herein, 在此处; 于此文件(或声明、事实)中
  164. However the assumption of rectified stereo images is very restrictive and, in many applications, is not suitable nor feasible 然而,这个双目相机已经校正的假设是非常受限的,而且在很多应用中都是不合适或不可行的。
  165. However, camera model abstraction raises some difficulties that need to be addressed, and are discussed next. 然而,相机模型的抽象化也带来了一些困难,这些在之后会提出并进行讨论。
  166. However, further questions need to be addressed in future work. 然而,在未来的工作中,更多的问题等待解决。
  167. However, the big leap in robustness is obtained by our novel system. 然而,我们的新系统在鲁棒性上取得了巨大的进步与提升。
  168. However, the trained models cannot be applied for arbitrary configurations, and re-training is often more resource intensive than applying a conventional calibration approach. In addition, the failure case analysis and analytical estimation of the limits of operations are highly challenging for black box deep learning approaches. 然而这些训练好的模型并不能适用于任意的硬件配置环境,并且重新训练一个网络通常比应用常规校准方法更耗费资源。此外,对于黑箱的深度学习方法来说,故障案例分析和操作限制的分析估计是非常具有挑战性的。
  169. If a camera is the only sensor used, the system has no way of knowing scale. It is impossible with only one image to tell the scale of an object. 如果摄像机是唯一使用的传感器,则系统无法知道尺度信息。只有一幅图像是不可能分辨出物体的尺度大小的。
  170. If any of these assumptions are violated 如果这些假设中有任意一条被违反了
  171. Image matching is a prerequisite for image registration and the quality of tie points influences the registration accuracy. 影像匹配是影像配准的先决条件,同时同名点的精度也会影响配准精度。
  172. In a nutshell 简明扼要地说,简而言之
  173. In fact, it forms the basis for the subsequent processing. 事实上,它构成了后续处理的基础。
  174. In order to accomplish this accurately and efficiently 为了精确与高效地完成...
  175. In order to localize, one needs a map. In order to create a map, one must localize. 为了定位,需要一张地图,而要创建地图,又必须要定位。
  176. In order to satisify regulatory requirements... 为了满足监管要求...
  177. In our case 在我们的情况中
  178. In our experiments, we observed a common problem that... 在我们的实验中,我们观察到了一个场景的问题...
  179. In the absence of groundtruth poses, we compare our estimates against that of ORB-SLAM2. 由于真实位姿的缺失,我们将我们估计的结果与ORB-SLAM2进行了对比。
  180. In this way, simultaneous map building and localization can be seen to present a question of which came first. 这样,同时的地图构建与定位就把“谁先来”这个问题摆在了我们面前
  181. In this work, we conducted a survey of ... 在这项工作中,我们对...进行了调查
  182. In words, xxx aims to find the x that minimizes the sum of covariance weighted visual and inertial residuals. 也就是说,xxx的目标是找到最小化协方差加权视觉和惯性数据残差之和的x。
  183. Inequality holds for any weighted mean of the two points. 不等关系对于任意两个点的加权平均仍然成立。
  184. It implies that the image-matching based on grey correlation is sensitive to the image grey feature, which indicates that the more similar the reffered band with the test band, the better the evaluation results. 它说明了基于灰度相关的影像匹配对于影像的灰度特征很敏感,也就是说参考波段与待配准波段越相似,评价的结果就越好。
  185. It is difficult for one method to work for all cases. 一个方法很难在所有场景下工作。
  186. It is important that sensors used be low cost with simple calibration requirements. 重要的是,使用的传感器应该使用成本低且校准要求简单。
  187. It is important to develop methods that take full advantage of this information. 开发一种可以完全利用这个信息的方法是非常重要的。
  188. It is reasonable to believe that 有理由相信
  189. It is trivial for humans to make this distinction. 对于人类而言做这种判别是很简单(微不足道)的。
  190. It is worth nothing that ... 值得注意的是
  191. It provides an alternative way for image registration to conventional orthorectification with an accurate digital elevation model and known sensor geometry, both of which are sometimes difficult and complex to acquire. 相比于需要精确数字高程模型和已知的传感器几何参数(这两种数据有时候很难获取或很麻烦/复杂才能获取到)的传统正射校正方法,它为影像配准提供了一种替代的方法。
  192. It remains future work to automatically select the minimal number of basis functions for a given motion. 在给定一个运动的情况下,自动选择基函数的最小个数仍然是未来需要做的工作。
  193. It usually suffers a drop in localization accuracy when ... 当...的时候,通常在定位精度上会有下降
  194. L0 through L6 从L0到L6
  195. Landmarks are detected and compared with a map to localize the robot while at the same time newly detected landmarks are added to that map. 检测地标并与地图进行比较以定位机器人,同时将新检测到的地标添加到地图中。
  196. Lights can grow in intensity without increasing in extent and vice versa. 灯光可以在不改变范围的情况下增加亮度,反之亦然。
  197. M is of very large scale M的尺寸非常大
  198. Methodology Taxonomy 方法分类
  199. Multivariate Gaussian Model 多变量高斯模型
  200. NTL research is quickly advancing NTL研究正在快速发展
  201. Nonlinear Least Square(NLLS) problem 非线性最小二乘问题
  202. Nonlinear image registrations need to be exploited instead. 非线性影像配准需要被研究。
  203. Note that the standard deviation is quite high. 注意标准差非常大。
  204. Numerous image registration techniques have been proposed in the literature. 许多影像配准技术都被提出了。
  205. ORB is at two orders of magnitude faster than SIFT ORB比SIFT快了两个数量级
  206. ORB-SLAM3 achieves in all sensor configurations more accurate result than the best systems available in the literature, in most cases by a wide margin. ORB-SLAM3在所有传感器配置中都取得了比现有最好方法更精确的结果,在大部分情况下好一大截。
  207. ORB-SLAM3 more than doubles the accuracy of VI-DSO. ORB-SLAM3是VI-DSO精度的两倍多。
  208. Objects are embedded in heavy clutter and vary significantly in pose and appearance. 物体都严重地聚集同时在姿态和外观上都有显著变化。
  209. On the other hand our SVS fully exploits the fact that... 另一方面,我们的SVS方法充分证实了...这样一个事实
  210. One can see that 可以发现
  211. One major drawback of the above approaches is that 上述方法的一个主要缺点是
  212. Only keep in mind that ... 唯一需要记住的是...
  213. Our objective is to seek an appropriate T that is accurate enough to model the scene deformation. 我们的目标就是寻找到一个合适的T,使得其可以精确建模场景畸变。
  214. Over the past years, numerous researchers have been devoted to automating crack detection. 在过去几年中,大量的研究者都投入到了自动裂缝检测中。
  215. Overall, feature-based methods are more robust to the geometric distortions of images compared with area-based methods, whereas area-based methods have better resistance to the non-linear intensity differences among images. 总体而言,相比于基于区域的方法,基于特征点的方法对于影像的几何变形更加鲁棒,同时,相比于特征点法,基于区域的方法对于影像间灰度的非线性变化有更好的不变性。
  216. PR curve PR曲线
  217. Prior to 1992 在1992年之前
  218. Products have a pointing accuracy of a 5 meters CE90, meaning that 90 percent of all WorldView Basic and Standard products will achieve a horizontal accuracy of 5 meters or better. 产品的指向精度为5米CE90,这意味着90%的WorldView基本和标准产品将达到5米或更好的平面精度。
  219. Push broom hyperspectral imaging is a process where a number of 1-D linear images are continuously sampled from a line-scan camera. 推扫高光谱成像就是从一个线阵相机中连续不断地采样,得到一维影像。
  220. Recent work focuses primarily on ... 最近的工作主要关注...
  221. Relative Error(RE) 相对误差
  222. Run each test against all training patches. 在所有训练影像块上运行测试。
  223. SLAM techniques build a map of an unknown environment and localize the sensor in the map with a strong focus on real-time operation. SLAM技术关注实时性地在一个未知环境中建立地图并且确定传感器自身在这个地图中的位置。
  224. Several issues can be addressed by the use of large field-of-view cameras. 几个问题都可以通过使用大视场角相机解决。
  225. So far, it is safe to say that there is no single localization system that can work robustly for all applications. 目前,可以相对安全地说没有一个定位系统可以在所有应用场景下鲁棒地运行。
  226. Specific attention is given to ... 我们特别关注了...
  227. Such data pose steep challenges to existing vision algorithms 这些数据将巨大/艰巨的挑战摆在了已有视觉算法的面前。
  228. T is the transformation from the coordinate frame of the calibration grid to the camera coordinate frame for image n. T是从校正格网坐标系到第n个影像的相机坐标系的变换。
  229. The SIFT detector generates 67 keypoints in I1 and matched points in I2 respectively. SIFT分别在I1中得到了67个特征点,在I2中得到了67个匹配点。
  230. The WorldView-1 horizontal accuracy versus time is provided in Figure 3. WorldView-1平面精度与时间的关系如图3所示。
  231. The above transformation estimation can be defined as a finite search problem by discretization of the possible parameter space into equal bins. 通过对可能的参数空间进行离散化成相等的区域,上面的变换估计可以被定义为一个有限搜索问题。
  232. The affine transformation matrix between frames does not account for the roll and pitch of the camera out of the image plane. 帧之间的仿射变换矩阵不考虑相机相对于图像平面的滚动和俯仰。
  233. The block-based scheme evenly distributes the interest points over the image. 分块策略可以平均分布影像中的特征点。
  234. The comparison against the state of the art shows that DynaSLAM achieves in most cases the highest accuracy. 与最新的SLAM比较表明,DynaSLAM在大多数情况下达到了最高的精度。
  235. The continuity is better as CI gets closer to 1. 当CI越接近于1的时候,连续性就会越好。 
  236. The correspondence points between the reference and sensed image. 参考影像与拍摄影像间的对应点。
  237. The error between the location of the projected pixel and the GCP is the error for that point.  投影像素的位置和GCP之间的差异就是该点的误差。
  238. The evaluation results demonstrate that the model based on high-order polynomials (i.e., KB 8) shows the lowest reprojection error but is 5-10 times slower than competing models.  结果显示,基于高阶多项式的模型(如KB 8)拥有最小的投影误差,但相比于其它对比算法慢了大约5到10倍。
  239. The failures of these methods can be attributed to ... 这些方法的失败可以归结于...
  240. The ground control points are used for accuracy testing purposes and are not used to correct the position of the image. The difference in the location of feature representing the control point in the image, and actual location of the control point is measured in the image. This difference is the error for that pixel. The errors for all the control points in the image are measured and these errors are used to calculate the accuracy of the image.  地面控制点仅用于精度评价,并不用于校正图像位置。控制点的实际位置以及影像中表示该控制点的特征位置之间的差异被测量,这个差异就作为该像素的定位误差。最后图像中所有控制点的误差都被测量,基于这些误差来计算图像的精度。
  241. The high barrier to entry prevents many students from taking these courses and being exposed to the field. 很高的入门门槛阻碍了学生选择这些课程以及了解该领域。
  242. The horizontal and vertical axes are error in the East and North directions, respectively, both in units of meters. 水平与竖直坐标轴分别表示东和北方向上的误差,单位都是米。
  243. The large performance gap between ours and ORB-SLAM (short) suggests that 我们方法与ORB-SLAM方法巨大的表现鸿沟(差异)说明了
  244. The majority of robots still lack even the most basic forms of autonomy. 大多数机器人甚至还缺乏最基本的自动形式。
  245. The morphological filter is fairly simple yet good enough to refine binary masks to properly overcover the clouds and their shadows. 形态学过滤虽然相当简单,但是对于精化二值掩膜却有着很好地效果。
  246. The motivation of this paper is to address this deficiency by performing a comprehensive evaluation of publicly-available VIO algorithms on hardware configurations that are typical of flying robot systems. 本文的目的是通过在飞行机器人系统的典型硬件配置上对公开可用的VIO算法进行综合评估来解决这一缺陷。
  247. The narrower viewpoints limit the possibilities of a general robust solution. 较窄的视角限制了一个一般的、鲁棒的解决方案的可能性。
  248. The outline of the paper is the following 论文的提纲如下
  249. The pipeline could be sped up considerably by not matching all the query descriptors to the training data but our goal was only to show the feasibility of object detection with ORB. 这个流程可以通过不匹配所有训练数据来大幅提升效率,但我们这里的目的只是为了展示ORB在物体检测中的可行性(所以没有这么做)。
  250. The position shifts are within 3 pixels. 位置漂移在3个像素以内。
  251. The reason for this 这个的原因是
  252. The registration result is further fine-tuned in the second stage by a point-to-point dense match based on a highly accurate optical flow estimation. 在第二阶段,配准结果会进一步精化,通过一个点对点的稠密光流匹配和估计。
  253. The relative distance is increased in steps of 0.1 up to 3.0. 相对距离以0.1的步长增加到3.0。
  254. The remainder of the paper is organized as follows. 文章剩余部分按如下组织。
  255. The remainder of this paper is organized as follows. 本文的剩余部分按如下组织。 
  256. The remainder of this paper will present the details of each of these principles. 本文的其余部分将详细介绍这些原则。
  257. The reminder of this paper is organized as follows. 文章的剩余部分按如下组织。
  258. The rest of the letter is structured as follows ... 这篇letter的其余部分结构如下...
  259. The rest of the tutorial is structed as follows. 本教程的其余部分结构如下。
  260. The rest of this thesis is dedicated to explaining the details ... 这篇论文的其余部分致力于解释...的细节
  261. The results demonstrate that with similar accuracy, our model shows around five times faster computation time for the projection function than the KB 8 model. 结果显示,在类似的精度下,我们的算法比KB8模型在投影函数的计算上快了大约5倍。
  262. The simplicity and efficiency of our method, in combination with its tendency to yield unexpectedly high recognition rates on challenging data, could make it a good baseline for “calibrating” new datasets. 我们方法的简单与高效,以及在挑战性场景下的高识别率,可以让它成为一个“校正”新数据集的基准。
  263. The subscript i indicates... 下标i表示...
  264. The system acquires multiple images of the same location on the ground while collecting data in a "push-frame" approach. 该系统采用“推帧”方式采集数据,同时采集地面同一位置的多幅图像。
  265. The test is performed in the following manner 测试按照下面的方式进行
  266. There is no free lunch in visual state estimation. 在视觉状态估计中没有免费的午餐。
  267. Therefore, a hybrid model is likely to be used to strike a balance among accuracy, implementation cost and computational cost. 因此,我们可能使用一个混合模型来寻找精度、实现代价和计算代价之间的平衡。
  268. These developments prompt us to revisit the problem of robustly aligning batches of linearly correlated images. 这些进步与发展促使我们重新研究批量线性相关影像的鲁棒对齐这个问题。
  269. These errors are large in magnitude, but sparse in the spatial domain. 这些误差在振幅上很大,但是在空间域上很稀疏。
  270. These limitations block the evaluation and improvement of methods. 这些限制了方法的评估与进步。
  271. These nighttime lights (NTL) can be resolved from space by satellites. 这些夜晚的灯光可以通过太空中的卫星捕捉到。
  272. These results are particularly encouraging 这些结果尤其令人鼓舞
  273. They are the key of the better accuracy. 它们是更高精度的关键。
  274. They paint an accurate picture of obstacles faced by real, situated robots. 他们准确地描绘了真实的情形下机器人所面临的障碍。
  275. This difference should be kept in mind when comparing data from the two satellites. 当比较两个卫星数据的时候,应该时刻记得这个差异。
  276. This improvement comes at the expense of introducing ... 这个提升带来的代价是引入了...
  277. This is especially problematic for studies of ... 这对于...的研究尤其是个问题
  278. This is obviously not the case if we use the difference as an error metric directly. 如果我们直接使用差分作为误差度量,显然不是我们期望的结果。
  279. This iteration converges to a local optimum. 这个迭代收敛到局部极值。
  280. This limitation can partially be overcome by ... 这个限制可以被...部分地克服
  281. This opens the door to a range of studies using multitemporal NTL imagery, including ... 这为很多利用多时相夜光影像的研究打开了一扇门,包括...
  282. This phenomenon significantly reduced the performance and robustness of the matching. To handle this issue we introduce a pre-processing step 这个现象显著降低了匹配的性能与鲁棒性。为了解决这个问题,我们引入一个预处理步骤...
  283. This section outlines the stages of this program. 本部分列出了该程序的主要阶段。
  284. This survey of available systems shows that existing platforms do not satisfy all of the educational goals. 对现有系统的调查表明,现有的平台并不能满足所有的教育目标。
  285. This variation complicates efforts to estimate regional or global population levels from NTL. 这个差异使得利用夜光数据估计区域或全球人口变得复杂。
  286. Though semi-automatic methods may yield very accurate calibration, these approaches are very time consuming and the calibration results highly depend on the skills of the operators. 虽然半自动方法可能产生非常精确的校准结果,但这些方法非常耗时,校准结果高度依赖于操作员的技能。
  287. Throughout the paper 全篇文章/论文
  288. Time synchronization of sensors is a crucial aspect of building a robotic system. 传感器的时间同步是构建一个机器人系统十分重要的方面。
  289. Timing for 10000 operations in micro seconds measured on Intel Xeon E5-1620. 以毫秒表示的在Intel Xeon E5-1620上运行10000次操作的耗时。
  290. To accommodate the vehicles in small size in satellite videos 为了容纳/适应卫星视频中较小尺寸的车辆 
  291. To address the first problem 为了说明第一个问题
  292. To address this problem, a block-based scheme, which is free of threshold, is adopted in the extraction of interest points. 为了解决这个问题,在提取特征点时采用了一种基于分块的策略,这避免了设置阈值。
  293. To address this problem, a common solution is to ... 为了解决这个问题,一个通常的做法/解决办法是...
  294. To benefit the reproducibility of related research, we release our implementation of the methods in this tutorial to the public. 为了方便相关研究的重复工作,我们将这篇文章中我们方法的实现(代码)公开了。
  295. To bridge this gap in performance on real images, we developed 为了填补在真实影像上性能表现的鸿沟,我们开发了
  296. To deal with the aforementioned issue, we introduce context information to lower-level layers to generate a feature pyramid through a top-down architecture, which is inspired by [43]. 受[43]的启发,为了解决前面提到的问题,我们通过一个自上而下的结构,将上下文信息一层一层地引入并生成一个特征金字塔。
  297. To demonstrate the superiority and generalizability of the proposed method, 为了演示我们提出方法的优越性和通用性
  298. To get robustness against this situation 为了在这个场景中获得鲁棒性
  299. To handle the merging problem of... 为了解决这个出现的...问题
  300. To improve the rotation invariance 为了提高旋转不变性
  301. To our best knowledge 据我们所知
  302. To overcome the aforementioned shortcomings, 为了克服前面提到的不足
  303. To quantify the improvements of... 为了量化...的提升
  304. To remedy this delay, 为了缓解这种延迟 
  305. To simplify notation, we use A instead of B in the remainder of this section. 为了简化标记,在剩余部分,我们使用A而不是B。
  306. To this end 为了达到这个目的
  307. To this end, ... 为了达到这个目的
  308. To this end, ... 为了这个目的
  309. To verify the effectiveness of S-LSD 为了验证S-LSD算法的有效性 
  310. Tracking performance still lags behind geometric approaches. 跟踪效果始终落后于几何方法。
  311. Two commonly used error metrics are the absolute trajectory error(ATE) and the relative error(RE). 两种常用的误差度量是绝对轨迹误差(ATE)和相对误差(RE)。
  312. Typically higher complexity usually means it is more power demanding, more infrastructure demanding, and more computation demanding. 通常更高的复杂度意味着更多的电源消耗、更多的硬件资源和更多的计算。
  313. Unfortunately, 不幸的是
  314. Univariate Gaussian Model 单变量高斯模型
  315. Unless otherwise stated, we always set 除非特别说明,我们一直...
  316. VIIRS is not superior to OLS in all aspects. VIIRRS并不是在所有方面都比OLS要好。
  317. VO which is rooted in geometry should not be closely coupled with appearance. 根植于几何的VO不应该紧密地与影像外观耦合起来。
  318. VO/VIO is able to provide robust state estimate for other tasks, such as control and planing, and therefore is widely used in robotic applications. VO/VIO能够为控制和规划等其它任务提供鲁棒的状态估计,因此在机器人应用中得到了广泛的应用。
  319. We are currently exploring this idea. 我们目前正在探索这个想法。
  320. We are following that to be consistent in notation. 我们遵循它以在表达记法上保持一致。
  321. We are highly interested in understanding how to better leverage the power of data-driven methods to build a robust, scalable, flexible and generalizeable localization pipeline. 我们对理解如何利用数据驱动方法的强大能力来建立一个鲁棒、可缩放的、灵活的、泛化性好的定位流程有着浓厚的兴趣。
  322. We can find strengths and limitations in both methods, hense the motivation for their combined use. 我们可以发现这两种方法的优点和局限性,这也是将它们结合使用的动机。
  323. We can see "raw" estimation from VINS-Mono is in a different reference frame as the groundtruth, and therefore cannot be directly compared. 我们可以看到VINS-Mono直接的估计结果与真实在不同的参考坐标系中,因此它们不能够直接比较。
  324. We first relax the highly nonconvex objective function to its convex surrogate. 我们首先将这个高度非凸的目标函数简化(转换)为它的凸替代形式。
  325. We publish the source code, not only for the benefit of the SLAM community, but with the aim of being an out-of-the-box SLAM solution for researchers in other fields. 我们公开源码,一方面是为了方便SLAM社区,另一方面是期望成为其它领域的一些研究者拿过来就能用的SLAM解决方案。
  326. We then linearize the nonlinear equality constraint, yielding a sequence of convex programs that can be solved efficiently via modern first-order optimization techniques. 我们然后将非线性的等式约束线性化,从而产生一系列的凸优化问题,进而可以通过现代的一阶优化方法高效求解。
  327. When a map is not available and building one is not feasible, it is impossible to know a robot's exact location, so a best guess pose estimate can be maintained along with a measure of uncertainty. 当地图不可用且构建地图不可行时,就不可能知道机器人的准确位置,因此我们可以通过测量不确定程度来维持一个最优的位姿估计猜测。
  328. With the arrival of stronger hardware and better machine-learning techniques 随着更强大的硬件与更好的机器学习技术的到来
  329. Years of research in geometric computer vision 在几何计算机视觉多年的研究
  330. Yet there is little agreement on ... 但是对于...并没有什么共识
  331. You may need to do sequential operation, where the result of each iteration is used by the subsequent iteration. 你可能需要做一些有顺序的操作,换句话说每一次迭代的结果都会被下一次迭代所使用。
  332. a critically important role 一个十分重要的角色
  333. a few code improvements 一些代码方面的改进
  334. a more sophisticated solution 一个更复杂的解决方案
  335. a number of drawbacks 一系列的问题/缺点
  336. a quarter of these studies 大约四分之一的研究
  337. a slight degradation in performance 在性能上的轻微降低
  338. ablation study 烧灼研究
  339. abnormal action detection 异常行为检测
  340. acceptable accuracy 可以接受的精度 
  341. accuracy specifications 精度参数(指标)
  342. accuracy stability over time 精度的时间稳定性
  343. accurate depth with detailed texture and color information can be obtained in parallel from the scene 携带纹理和颜色的精确的深度信息可以从场景中并行地获得
  344. achieves state-of-the-art accuracy 达到了目前最高的精度
  345. actually performs poorly 事实上表现比较糟糕
  346. adopt a structure from motion (SfM) method 采用SfM的方法
  347. aerospace-grade 航空级
  348. affine-invariant 仿射不变的
  349. all objects within a class have roughly the same size 同一类的所有物体都有差不多同样的尺寸
  350. analytical solution 解析解
  351. annual composites 年度合成产品
  352. apart from 除了
  353. applicability may be limited 应用可能会受限
  354. application domain 应用领域
  355. approaches can be methodologically divided into two main categories 这些方法可以主要被分为两类
  356. are consistent regardless of 尽管...,...是保持不变的(一致的)
  357. are copied verbatim from ... 从...一字不差地复制过来
  358. are incapable of 不能够...
  359. assess accuracy 精度评价
  360. assumes the presence of a flat ground 假设一个平地的存在
  361. assuming not all images are colinear 假设不是所有的影像都共线
  362. at the expense of 伴随着...的代价
  363. automatic road crack detection method is required to boost this progress 需要自动道路裂缝检测来加速这一过程
  364. available in the literature 文献中可以查阅到的
  365. baseline for comparison 用于比较的基准方法
  366. batch-processing 批处理
  367. be augmented with 受...的增强
  368. be augmented with ... 由...增强、支持
  369. be safely triangulated 被安全(准确)三角化
  370. best correspondence 最好的对应关系(匹配)
  371. blooming should be of particular concern when ... 当...的时候,尤其需要注意溢出效应
  372. bold uppercase letter 加粗大写字母
  373. borrows ideas from this literature 从这个文献中借鉴了想法
  374. both these algorithms share the same limitation that 这些算法都有相同的局限性...
  375. bridge the gap between ... and ... 架起了...和...的桥梁
  376. brighter and dimmer 明亮与暗淡
  377. broader spectrum of sensors and setups 更广泛的传感器与配置范围
  378. brute force descriptor matching against the keyframe 在关键帧间的暴力描述子匹配
  379. but this solution is obviously expensive 但显然这个方法的代价很高
  380. c.f. 对照,参看
  381. camera coordinate system 相机坐标系
  382. can be reliably extracted and matched across images 可以在影像间被可靠的提取与匹配
  383. canonical approaches 传统方法/经典方法 
  384. canonical model 经典模型
  385. carefully designed and specifically fine-tuned 仔细设计和特殊精调
  386. cheapest and smallest sensor setup 最便宜与小型的传感器配置
  387. chronologically 按时间序列发生地
  388. clearly visible 明显可见的
  389. collinearity equation 共线方程
  390. color composite 彩色合成 
  391. come up with something novel 想出一些创新的东西
  392. common image region 公共的(共有的、重合的)影像区域
  393. community transmission 社区传染,社区感染
  394. compelling results 令人满意/开心/信服的结果
  395. competitive performance 有竞争力的表现
  396. competitive result 很有竞争力的结果
  397. complementary information 互补的信息
  398. complexity is independent of the map size (计算)复杂度独立于地图尺寸
  399. compromise between computational cost and accuracy 计算成本与精度的折中
  400. computation expense 计算消耗/耗费
  401. computation times 计算耗时
  402. computational complexity 计算复杂度
  403. computational cost 计算代价/消耗
  404. computational efficiency 计算的高效率
  405. computational overload 计算负担
  406. computationally demanding 需要计算资源的、消耗计算资源的
  407. computationally efficient 计算高效地
  408. computationally expensive 耗费计算资源的
  409. computationally expensive trigonometric operations 计算量巨大的三角函数操作
  410. computationally tractable 可接受的计算
  411. computationally-efficient 计算高效地
  412. conduct the experiments 做实验
  413. consecutive frames 连续帧
  414. considerably promising 相当有前途的
  415. constellation of many satellites 多个卫星的星座
  416. consumer market 消费市场
  417. conveyor belt 传送带
  418. cope with ... (成功地) 对付,处理... 
  419. correctly scaled 正确缩放的
  420. cost-wise 成本导向
  421. counts the number of nonzero entries in the error matrix ...统计错误矩阵中非零元素的个数
  422. crack and edge detection share similar characteristics in shape and structure 裂缝和边缘检测在形状特征和结构上具有相似性
  423. crisply defined foreground regions 干脆的前景区域
  424. criterion is satisfied exactly 判别条件被精确满足
  425. cues they exploit 他们利用的线索
  426. cumulative errors 累积误差
  427. cumulative iteration 累加迭代 
  428. dead reckoning 推算定位;航位推测法
  429. decent performance 漂亮的性能/表现
  430. decoupled from 解构,分离
  431. demonstrated impressive levels of performance 展示了令人印象深刻的性能
  432. demonstrating higher accuracy and efficiency than existing methods 展示了比已有方法更高的精度与效率
  433. densely and evenly distributed over the whole image 均匀且稠密地分布在整个影像中
  434. detect when the sensor returns to a mapped area and correct the accumulated error in exploration 检测传感器什么时候回到已经建过图的区域以及校正探索过程中的累积误差
  435. dimensionality reduction 降维
  436. dimensionality-reduction method 数据降维方法
  437. discrete parameterization 离散参数化
  438. discriminative representations 区别性的表达
  439. divide and conquer 分而治之
  440. downlink&uplink 下传与上传
  441. downstream applications 后续的应用
  442. dramatic increase 巨大的增长
  443. drawn from 来自
  444. drive up 抬高
  445. due to persistence of thin clouds 由于薄云的存在
  446. dynamic changes in the motion and in the environment 运动与环境中的动态变化
  447. efficiently and reliably recover low-dimensional structures from high-dimensional data, despite gross observation errors 从包含有观测粗差的高维数据中高效、可靠地恢复低维结构
  448. ego-motion 自身运动
  449. empirically estimated coefficients 经验估计的参数
  450. eroding and dilating 腐蚀与膨胀 
  451. evaluate the method 评价这个方法
  452. evaluation criteria 评价标准/指标
  453. even though 即使
  454. even-distributed 均匀分布的
  455. exceed a threshold 超过阈值
  456. exhaustive search 耗费精力的搜索
  457. exhibit high initial eigenvalues 展示(展现)了很高的初始特征值
  458. existing research 现有研究
  459. expensive high quality sensors 昂贵的高质量传感器
  460. experimental results 实验结果
  461. experimental results 实验结果
  462. expert parameter tuning 专家级(专业)的参数调校
  463. extensive evaluation 大量的评价/测试
  464. extensive review 广泛的调研/总结
  465. faint trace 微弱的痕迹,蛛丝马迹
  466. fair and thorough performance measurement 公正彻底的性能衡量
  467. falls off/degrade sharply/dramatically 显著降低、下降
  468. fault tolerance 容错性
  469. favorable performance compared to other learning-based method 相比于其它基于学习的方法(取得了)令人满意的表现
  470. flip the signs 反转符号
  471. focal length 焦距
  472. for the first time in China 第一次在中国...
  473. formulate the optimization function as follows 将优化函数写成如下形式
  474. formulates batch image alignment as a rank minimization problem 将批量影像对齐建模成一个秩最小化问题
  475. fundamentally remove the problem 根本上解决这个问题
  476. further improved upon 进一步提高
  477. fuse data from multiple sensors 融合来自多传感器的数据
  478. generalization of those methods 这些方法的泛化性
  479. give depth information rapidly and accurately 快速准确地给出深度信息
  480. globally uniform and continuous measurement 全球统一和连续的测量/观测
  481. golden rule to follow 遵守的黄金法则
  482. gross error 粗差
  483. ground robots 地面机器人
  484. ground sample distance(GSD) 地表采样距离,对应中文的地表分辨率
  485. hand-crafted 人工设计的
  486. hand-crafted local descriptor 人工设计的局部描述子
  487. have been well studied in early days of computer vision 在计算机视觉领域的早期已经被很好地研究了
  488. heritage satellite 先前的卫星
  489. high accurate ground control points 高精度地面控制点
  490. high spatial resolution 高空间分辨率
  491. high temporal resolution 高时间分辨率
  492. high-technological companies 高科技公司
  493. higher spatial and radiometric resolution 更高的空间与辐射分辨率
  494. highly active field of research 高度活跃的研究领域
  495. highly corrupted data 严重污染/污损的数据
  496. highly dynamic scenes 高动态场景
  497. highly nonlinear 高度非线性
  498. histogram of gradient directions at 10 degree intervals 每10度为一个间隔的梯度直方图
  499. horizontal error/accuracy 平面误差/精度
  500. i.e. 也就是,亦即
  501. illumination change causing complex intensity variation 导致复杂灰度变化的光照变化
  502. image domain 影像域,影像坐标系
  503. image rectification 影像校正(根据变换模型重采)
  504. imagery 在论文中对应于“影像”的抽象概念,image是图片,像片的意思
  505. images covering the plain areas 覆盖平原地区的影像
  506. in accordance with 依照
  507. in essence 从本质上说
  508. in nature 性质上,实际上
  509. in some specific cases 在某些特殊场景
  510. in the literature 在相关文献中
  511. in-house (公司或机构) 内部存在的,内部进行的
  512. include but are not limited to 包括但不限于
  513. increasing popularity 越来越流行(的趋势)
  514. increasingly fine resolutions 越来越高的分辨率
  515. inherent inability to measure scale 测量尺度的先天能力缺陷
  516. inherited accuracy and generalization problems 内在的/天生的精度与泛化性问题
  517. insensitivity 不敏感性
  518. intense research 众多的(密集的)研究
  519. intensity inhomogeneity 灰度不均匀/不一致
  520. intensive search 密集的搜索
  521. intercalibration 交叉定标
  522. interdisciplinary collaborations 跨学科合作
  523. interesting line of research 有趣的研究线(方向)
  524. intractable to deal with 很难对付(或处理)的
  525. irrelevant correspondences 不相关的对应匹配
  526. is more liable to be 更有可能受…影响
  527. is partitioned into bins ...被划分成多个间隔
  528. is primarily driven by 主要由...影响
  529. it is worth noting that 值得强调的是,值得注意的是
  530. it should be emphasized that... 需要强调的是...
  531. iterate until convergence 迭代直到收敛
  532. iterative refining procedure 迭代精化的过程
  533. its reliance on 它对于...的依赖
  534. jointly estimating 联合地估计
  535. labor-intensive 劳动密集的 
  536. large computational burden 巨大的计算负担
  537. latent variable 隐变量
  538. lead to a breakthrough in 在...带来了突破
  539. leverageing on 依赖于,基于
  540. lightweight 轻量的
  541. line & sample 行、列
  542. linear push-broom camera 线阵推扫相机
  543. linear push-broom satellite 线阵推扫卫星
  544. lit pixels 明亮的像素
  545. longer time spans 更长的时间范围
  546. loosing the advantages 损失了优势
  547. low-dynamic scenes 低动态场景
  548. lowercase letter 小写字母
  549. main novelties 主要的创新点
  550. main system components 主要的系统组成(构件)
  551. manual collection of tie points 人工选择同名点
  552. manual crack detection is considerably tedious and requires domain expertises 人工裂缝检测相当枯燥乏味且需要行业的专家
  553. many chicken-egg problems in computer vision 计算机视觉中很多的鸡生蛋、蛋生鸡问题
  554. measure of cornerness 角点的度量
  555. mechanically complex 机械上很复杂的
  556. medium-to-large 中大型的、中等以上的
  557. methods to deal with noise and uncertainty 处理噪声与不确定的方法
  558. mimic this approach 效仿这个方法
  559. misalignment 未配准、未对齐
  560. missile-firing 导弹发射
  561. mobile agent 移动体
  562. model the background 给背景建模
  563. modification of pyramid match 金字塔匹配的修改、改进
  564. monotonically decreasing 单调递减
  565. more computationally intensive 更加的计算密集
  566. more robust performance 更鲁棒的性能
  567. most light sources fall under the saturated value 大多数的光源都在饱和值以下
  568. most of the options are well explored 绝大多数的可能性/配置都被尝试/探索过了
  569. most popular implementation choice 最流行的实现选择
  570. motion estimation pipeline performs better than or on par with ORB-SLAM2 运动估计流程比ORB-SLAM2表现更好或者不分上下
  571. moving objects take up only a tiny portion of the scene of a video 运动目标只占整个视频场景很小的一部分 
  572. much more severe 更加严重
  573. multi-modality 多模态
  574. multipath effect 多路径效应
  575. multiresolution 多分辨率的
  576. multitemporal analysis 多时序分析
  577. naturally corresponds to 天然对应于
  578. neighboring local features 相邻的局部特征
  579. noise introduced by ... 由...引入的噪声
  580. non-collinear correspondences 不共线对应关系
  581. non-convex optimization 非凸优化
  582. non-robust system 不鲁棒的系统
  583. not applicable 不适用的
  584. not discriminative enough to differentiate the cracks from complex background with low level cues 不足够在低级(影像)线索的复杂背景下来区分裂缝
  585. not fine enough 不够精确
  586. not well investigated in previous vision tasks 在以前的视觉任务中没有得到很好的研究
  587. notably better
  588. noteworthy 值得注意的
  589. numerical stability 数值稳定性
  590. observation generated by various sensors 由不同传感器得到的观测
  591. observed vectorized image frames 观测的序列的影像帧
  592. obtuse angle 钝角
  593. off-the-shelf 现成的,拿过来就可用的,市面上有的,能够买到的
  594. off-the-shelf 现货的、能买到的
  595. offers insights into 提供对...的洞察,深刻理解
  596. on-the-fly 在飞行过程中实时的,在线的
  597. only holds locally ...局部成立(满足条件)
  598. open question 开放的问题,悬而未决的问题,待解决的问题
  599. orbital dynamics 轨道动力学
  600. order the FAST keypoints 对FAST角点排序
  601. ordinary least-squares regression model 普通的最小二乘回归模型
  602. ordinary people 普通人
  603. original image 原始影像
  604. originally designed for ... 原本设计用来...
  605. our method outperforms the alternatives 我们的方法比其它方法表现更好 
  606. our studies reveal that ... 我们的研究显示...
  607. outperform ... by a margin 比...好了一大截
  608. over existing alternatives 比已有方法...
  609. overly restrictive assumptions 过分苛刻的假设
  610. parallel and distributed computing 并行与分布计算
  611. parameter tuning 参数调整,调参
  612. partial occlusion 部分遮挡
  613. partially eliminate noises and enhance continuity of detected cracks 部分消除了噪声并且增强了检测到的裂缝的连续性
  614. particularly troublesome 尤其麻烦
  615. partition the image into subblocks and compute histograms (or histogram statistics, such as means) of local features in these subblocks 将影像分割成小块并在每块中计算局部特征的直方图
  616. partitioning the image into increasingly fine sub-regions 将影像分割成越来越细的子区域
  617. pavement surface condition 路表状态
  618. performance drops considerably 性能显著降低
  619. pesticide-spraying 农药喷洒
  620. photoelectron integration time 光电子积分时间
  621. pixelwise accuracy 像素级精度
  622. pointing capability and pointing stability 指向能力与指向稳定性
  623. polar-orbiting 极轨
  624. poorly triangulated visual features 三角化不良的视觉特征
  625. pose estimation accuracy 位姿估计精度
  626. posterior probability distribution 后验概率分布
  627. potential threat 潜在的威胁
  628. potentially key source 潜在的关键资源
  629. predefined tolerance 预先定义好的容忍度
  630. preliminary version 初步的版本
  631. prerequisite step 必要的前序步骤
  632. presence of obstacles 障碍的存在
  633. presents a challenge as well as an opportunity to many areas 展示了挑战,同时对于很多领域来说也是机遇
  634. prior to using any solution 在使用任何方法之前
  635. problem-specific 具体问题的
  636. problems we encountered are described in ... 在...描述了我们所遇到的问题
  637. program them by hand 人工编程
  638. projective nature of cameras 相机的投影特性
  639. promising results 有说服力的、可信的结果
  640. promising results 有希望的/不错的结果 
  641. provide insight into 对...的洞察、发现
  642. provided of course that 假设
  643. published results 已出版/发布的结果
  644. qualitative accuracy 定性的精度
  645. qualitatively validate 定性地评价
  646. quantitative evaluation 定量评价
  647. radiometric characteristics 辐射特性
  648. raise the threshold and try again 提高(增加)阈值,再次尝试
  649. rare device or system that is not widely used 没有被广泛使用的、少见的设备或系统
  650. raw camera and Lidar sensor streams 原始相机与Lidar数据流
  651. re could be required to reduce these distortions. 
  652. realistic environment 实际环境,真实环境
  653. realistic scenarios 实际场景
  654. reference system 参考系统(坐标系)
  655. registered image 配准的影像
  656. registration error 配准误差
  657. reject edges and provide a reasonable score 过滤边缘点并提供一个合理的分数
  658. reliable information 可靠的信息
  659. relieve this scale problem 减轻这个尺度问题
  660. remains a challenge 仍然是一个挑战
  661. representative results 代表性结果
  662. represented by ... 由...表示
  663. resource constained system 资源受限的系统
  664. restricted computation resource 有限的计算资源
  665. rigorous corresponding relationship 严密的对应关系
  666. robustness to lighting, blur, and perspective distortion 对于光照、模糊以及投影变形的鲁棒性
  667. root of this problem lies in that 这个问题的关键在于
  668. rotational invariance 旋转不变性
  669. rural-urban transition zone 城乡结合部
  670. salient points matching 静态点匹配
  671. sample training images 样本训练影像
  672. saturation prohibits the detection of variation in lighting levels beyond DN=63 过饱和阻碍了对于超过63的亮度等级变化情况的探测
  673. scale ambiguity 尺度不确定性
  674. scenarios with low illumination and motion blur which are among the main reasons causing failures of traditional purely vision-based VO systems 在一些场景中存在低光照或运动模糊,而这些是导致传统纯基于视觉的VO系统失败的主要原因
  675. scene appearance change across different frames is dominated by the camera motion 不同帧之间场景外观上的变化主要由相机运动引起(而并非由场景内部的运动引起)
  676. seasonal periodic noise 季节性的周期噪声
  677. second-best 第二好的
  678. self-contained 齐全的,完整的,完善的
  679. sensor degradation 传感器退化,老化
  680. sensor uncertainties 传感器不确定性
  681. set at a value 设为...值
  682. severely limited descriptive ability 严重限制了描述能力
  683. show competitive performance to state-of-the-art methods 展示了与最新方法相比很有竞争力的表现
  684. significant burden 显著的负担
  685. significant computational expense 显著的计算(资源)开销
  686. significant illumination variation 显著的光照变化
  687. significant progress 显著的进步
  688. significantly improved performance 显著提升的性能、表现
  689. similar results that are difficult to distinguish by the human eye 对于人眼很难辨别的相似的结果
  690. simple and computationally efficient 简单以及高计算效率
  691. simple but effective 简单却有效
  692. simplicity and eficiency 简洁与高效 
  693. single modality system 单模态系统
  694. slightly better 稍微好于...
  695. smoothness constraint 平滑约束
  696. solely depend on 仅仅依赖于
  697. some works have been devoted to ... 一些工作也投入在...
  698. somewhat larger 有些大
  699. spatial displacements 空间位移
  700. spatial distribution 空间分布 
  701. spatially coherent clusters 空间上有逻辑关联的一团
  702. spatially piecewise contiguous 空间上连续的
  703. spatiotemporal consistency 时空一致性
  704. spin off (从现有公司中) 剥离出,独立,分立
  705. spurred by 激励; 刺激; 鼓舞; 促进,加速,刺激(某事发生)
  706. standard PC 标准的电脑
  707. standard deviation 标准差
  708. static scenarios 静态场景
  709. status quo 现状
  710. still image 静态影像(相较于动态卫星视频) 
  711. stopping criterion (迭代的)停止判别条件
  712. stunning performance 令人深刻印象的性能
  713. sub-meter GSD 亚米级地表分辨率
  714. sub-pixel accuracy 亚像素精度
  715. subdivision scheme 分割原则、划分依据、方案
  716. subsequently 随后,接下来
  717. sudden illumination change 突然的光照变化
  718. sum synchronous orbit 太阳同步轨道
  719. sun-synchronous 太阳同步的
  720. superior performance 卓越的性能/表现
  721. surprisingly effective 令人惊讶的有效
  722. swath width (卫星条带影像的)幅宽
  723. take into account 考虑
  724. technically challenging 在技术上有挑战性的
  725. terrain-induced offset 地形引起的偏移
  726. textbook-level 教科书级
  727. the accuracy was consistent regardless of the platforms 无论平台如何,精度都是一致的
  728. the car is highly contrasting with the rest of the scene 相比于场景中的其它地方,汽车具有高对比度
  729. the challenges and complexity associated with robotics in the real world 现实世界中与机器人相关的挑战和复杂性
  730. the choice of threshold may significantly affect the outcome 阈值的选择会显著影响输出结果
  731. the higher value means the better performance 更高的值表示更好的表现
  732. the intensity threshold between the center pixel and those in a circular ring about the center 中心像素与其周围圆圈的像素的灰度(差异)阈值
  733. the lens will be covered with specks of dust 镜头会被灰尘颗粒覆盖
  734. the sensor cannot measure brightness levels beyond a value of 63 DN 传感器无法测量DN值大于63的亮度
  735. their inclusion can introduce inaccuracies 对它们的包含可能会引入误差,使得精度降低
  736. their measurements have highly diverse characteristics, which makes the efficient exploitation of the multimodal data challenging 它们的观测数据特性差异很大,这使得有效发掘这些数据成为挑战
  737. they are subject to a number of basic limitations 它们受限于一系列的基本限制
  738. think boldly 大胆想象、大胆思考
  739. this is not always the case 并不一直都是这样
  740. this practice is sufficient to deal with ... 这种(通用的、习惯的)做法对于解决...是足够的了
  741. this step may have a direct impact on the quality of the object based alignment algorithm 这个步骤可能对于基于目标的配准算法质量有直接的影响
  742. thorough evaluation 彻底的评价
  743. thorough understanding 彻底的理解
  744. three critical problems 三个关键的问题
  745. time series analysis 时间序列分析
  746. time-delay integration(TDI) CCD array 时延积分CCD阵列
  747. time-synchronized 时间同步的
  748. timing results 耗时测试结果
  749. to a certain extent 在某种程度上
  750. to the best of our knowledge 据我们所知
  751. top N points 前N个点
  752. trade-offs between error and computational resources 错误和计算资源之间的权衡
  753. trades off the rank of the solution versus the sparsity of the error 平衡结果的秩与错误的稀疏性
  754. train and test splits 训练与测试的划分
  755. training regime 训练策略,环境
  756. tremendous advances 巨大的进步
  757. tune 调整,调参等,tune parameters
  758. tuning the number of octaves 调整金字塔的层数
  759. uncover relationships between ... 揭示...的关系
  760. unprecedented success 前所未有的成功
  761. unseen scenes 没有见过的场景、没有去过的场景
  762. up to a third of pixels 至多三分之一的像素
  763. urgent need 迫切的需求
  764. usually does not pay off 通常都不成功
  765. variants or extensions of... ...的变种或延伸扩展
  766. various combinations of algorithms and sensors 不同算法与传感器的组合
  767. vertical accuracy performance meets the 5 meter accuracy specification 垂直精度性能满足5米精度规范
  768. vertical error/accuracy 高程误差/精度
  769. very costly 非常耗费资源的,非常耗时的,代价非常大的
  770. vice versa 反之亦然
  771. vice versa 反之亦然
  772. video surveillance 视频监控
  773. videos over a large territory 覆盖广阔土地范围 
  774. vis-a-vis 关于; 对于; 和…相比; 与…相较
  775. visual interpretation 视觉解译
  776. visually appealing solutions 看起来非常棒的方案
  777. visually striking 看起来很引人注目的
  778. vital role 关键角色
  779. w.r.t. with respect to,相对于
  780. was cropped at varying angles 以不同的角度进行截取
  781. we address the extrinsic calibration problem as a registration task 我们将外参标定问题作为一个配准任务
  782. we borrow from his centroid technique 我们借鉴他的质心技术
  783. we crop each image into 16 non-overlapped image regions 我们将每一张影像裁剪成16个不重叠的影像区域
  784. we discovered a large performance gap in our favor. 我们在我们的评价方式上发现了巨大的性能差异。
  785. we divide them into two sections 我们将它分为两个部分
  786. we exchange them with their SIFT counterparts 我们将它们与SIFT中的对应部分进行替换/交换
  787. we have found that this is rarely the case 我们发现这几乎不可能发生
  788. we need to impose that 我们需要推行/采用的是
  789. we terminate the iteration when 当...的时候,我们停止迭代
  790. well-aligned grayscale images 已经对齐的灰度影像
  791. widely available 广泛可用的
  792. will be valuable for new applications 对于新的应用是很有价值的 
  793. with the advent of deep learning 随着深度学习的到来
  794. without using any additional information about the spatial structure of the error 没有利用任何额外的有关于误差的空间结构信息
  795. work in the presence of ... 在...存在的情况下工作
↑ return to top ↑