کد مقاله | کد نشریه | سال انتشار | مقاله انگلیسی | نسخه تمام متن |
---|---|---|---|---|
531235 | 869820 | 2011 | 13 صفحه PDF | دانلود رایگان |
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.
Research highlights
► Taxonomy of correlation measures for stereo-matching.
► Evaluation protocol designed for correlation-based matching comparisons.
► Best measure near occlusions: Smooth Median Powered Deviation.
► Best measure in non-occluded areas: CENSUS.
► Rules for choosing the measure adapted to the constraints of a given application.
Journal: Pattern Recognition - Volume 44, Issue 9, September 2011, Pages 2063–2075