摘要： 提出了一种基于二维直方图和最大熵门限化的直线边缘检测方法。为了得到与实际边缘对应的单宽度直线,还利用了细化、连接与直线跟踪等处理方法,不但可以自动确定阈值,而且可以较好地保留弱边缘,去除噪声。最后得到单宽度的直线边缘,这对于目标识别和图像理解有着重要的意义。

关键词: 二维直方图, 最大熵门限化, 最大邻域差, 直线边缘

Abstract: A line edge detection algorithm is proposed in this paper,which is based on 2-D histogram and maximum entropy thresholding. After thresholding, the edges are thinned and edge points are linked to line fragments. Through the process of line tracing, the fragments are linked to longer lines. Final result is a single-width lines image. This algorithm has the feature that not only the threshold can be selected automatically, but also the weak edges are detected correctly most of the time and the noises are decreased considerably. The result line edge image is significant for object recognition and image understanding. ;;;

中图分类号: 

• TP391.41