Similarity measures for image matching despite occlusions in stereo vision
Similarity measures for image matching despite occlusions in stereo vision
复制标题
DOI：
10.1016/j.patcog.2011.02.001
复制
发表时间：
2011-09-01
影响因子：
8
通讯作者：
Crouzil, Alain
中科院分区：
文献类型：
--
作者：
Chambon, Sylvie;Crouzil, Alain
在计算机视觉的背景下，可以通过相似性度量来完成匹配。本文将这些措施分为五类。此外，还研究了先前为了处理遮挡问题而提出的基于稳健统计的 18 项措施，并与现有技术进行了比较。提出了一种新的评估协议和新的分析，结果强调了最有效的措施，首先是近遮挡、平滑中值动力偏差，其次是近不连续性、基于非参数变换的测量 CENSUS。 (C) 2011 Elsevier Ltd. 保留所有权利。
In the context of computer vision, matching can be done with similarity measures. This paper presents the classification of these measures into five families. In addition, 18 measures based on robust statistics, previously proposed in order to deal with the problem of occlusions, are studied and compared to the state of the art. A new evaluation protocol and new analyses are proposed and the results highlight the most efficient measures, first, near occlusions, the smooth median powered deviation, and second, near discontinuities, a non-parametric transform-based measure, CENSUS. (C) 2011 Elsevier Ltd. All rights reserved.