基于边缘方向信息的主动轮廓算法

EDGE DIRECTION BASED ACTIVE CONTOUR ALGORITHM

提出一种基于边缘方向信息的改进型主动轮廓算法ＭＡＣＡ（ＭｏｒｉｆｉｅｄＡｃｔｉｖｅＣｏｎｔｏｕｒＡｌｇｏｒｉｔｈｍ）．在Ｓｎａｋｅ模型的基础上加入２项基于匹配要求的新的能量函数，用ＦＧＡ（ＦａｓｔＧｒｅａｄｙＡｌｇｏｒｉｔｈｍ）迭代能量函数使其快速收敛．对轮廓曲线的一阶和二阶导数、搜索区域、轮廓的重抽样作锁定．

A method named MACA (Modified Active Contour Algorithm) was suggested.In the general Snake model, there is no required matching between orientation of the contour and edge direction in the gray level image.This kind of Snake model may fail in the images with rich texture or with concave objects.To avoid it, two energy functions based on the above mentioned ideas were proposed to improve the Snake model.To reduce the time cost in the iteration of MACA, a FGA(Fast Greedy Algorithm) was proposed. Also, the first order derivative and the second order, domain of searching area, contour resampling, and conditions of corner locking were redefined. Experimental results show that the proposed method is much better than the general Snake model,especially on searching for concave objects.