Curve Representation and Matching Based on Centroid Distance Increment Matrix
Yin Wen Dong · Jisuanji fangzhen · 2006
The extraction and recognition of boundaries from images are core problems in image understanding.In this paper,a new and quick method of irregular curve representation and matching is presented.This method does not represent and match every point of curve,rather,it samples the curve according to the principle of minimal area.Incremental distance between sample point and centroid of curve is used as status vector of sample point,then the centroid distance increment matrix is designed using these status vector of sample point.Lastly,it confirms the similarity of two curves by comparing the vectors of two curves.The experiment has shown that this method can not only exclusively represent curve,but also consumedly reduce calculation.A practical use of the proposed approach is to register a SAR image of certain region with maps.