Image Edge Detection Based on Relative Degree of Grey Incidence and Sobel Operator

摘要

Edge points are characterized by sharp transitions in gray levels in adjacent pixels, and relative degree of grey incidence can just reflect the degree of variations. In this paper an image edge detection method integrating relative degree of grey incidence with Sobel operator is presented. Firstly, the comparison sequence is constructed by sequentially ranking a certain pixel and its eight neighborhood of the image, and the reference sequence is formed by taking two orientation operator of Sobel operator. Secondly, the quantitative level difference between reference sequence and behavior sequence is decreased using initialization operation. Then the pixel concerned can be judged as an edge point when there exists a higher relative degree of grey incidence, which means similar geometric shapes of two sequences. By comparing the experimental results, it is proved that the strategy proposed in this paper can detect more details many traditional methods can not find.