摘要
本文介绍了基于字典的概率松弛标记法,将其用于边缘检测的后处理中,能充分利用像素间的关联信息,每次从一个像素的某一距离内考虑关联信息,这个距离将递增直到边缘特征被唯一确定。实验结果表明,该算法能获得连接性好、失真较少的精确的边缘图。
A new dictionary-based probability relaxation labeling algorithm is presented. As a post-processing method in edge detection, the proposed algorithm considers the label context within a certain distance from a pixel at a time. This distance keeps increasing until the label of a pixel is uniquely determined.
出处
《模式识别与人工智能》
EI
CSCD
北大核心
1999年第2期178-183,共6页
Pattern Recognition and Artificial Intelligence
关键词
边缘标记
概率松弛
结构字典
图像处理
Edge Labeling, Probability Relaxation, Configuration Dictionary
