图象角点检测的矢量场方法

Corner Detection Using Vector Potential

提出一种新的图象角点检测方法。该方法利用电磁场理论表达图象特征，进而从特征图象中提取出角点。Ｃａｎｎｙ边缘图象中的边缘被视为电流密度源而产生磁场。根据电磁场理论，可计算出相应的矢势。提出图象的角点位于该矢势图的鞍点和基于模板匹配的鞍点检测算法，并给出实际的角点检测结果。

This paper present a new corner detection method.Image features are represented using the theory of electromagnetics.Corners are detected in the vector potential of the field.Canny edge map is used as elemetary current density to create electromagenetic field.The vector potential of the field is calculated.We observed that corners are the saddle points in the vector potential.The saddle point detection algorithm is based on a simple template matching.The corner detection result is given.