Image correspondence based on region hierarchy
Y.-L. Chen, Hiromasa Nakatani · 2002
Finding correspondences of pictorial features is a basic problem in many areas, like map matching, looking up image data, or stereo vision. Region-based correspondence techniques seem to be effective for identifying the corresponding features, as regions are usually fewer than other image features and provide additional information such as color, area, and adjacency. This paper proposes to give each region a classified table of its surrounding regions, where all the regions are sorted in terms of their distances. Similarity between regions is computed by taking account of similarity of the corresponding classified tables. Then, one can determine correspondences of regions that provide the best resemblance between their classified tables.>