Gray-scale Edge Detection and Image Segmentation Algorithm Based on Mean Shift

Abstract

To solve the problem of the inaccurate segmentation for the gray image, a modified algorithm based on the mean shift is introduced. The modified algorithm constructs a novel kernel function histogram by combing the position information and the gray-scale information of a pixel, and then makes use of the mean shift algorithm with this new kernel function histogram to automatically detect the modes in the gray-scale image, which could be constructed fully by the kernel function defined above, filter and segment the gray image. Experiments based on a gray image with ground background are carried out by Canny, Sobel and the proposed mean shift method, and the results show that the mean shift algorithm could effectively extract not only bright object but also weak object, and the result of the introduced algorithm is more fit factual scene than that of the usual segmentation algorithm such as Canny and Sobel algorithm. DOI: http://dx.doi.org/10.11591/telkomnika.v11i3.2223 Full Text: PDF