关键词: 非参数变点统计分析, 边缘检测, 计算机视觉

Abstract:  Statistical method is one of important methods on edge detection of computer vision. This paper proposes a method of edge detecting which is based on the change point statistic analysis because it minimises the need for a priori information about images. What’s more, some problems for gray level images is studied to which Gaussian white noise and salt and pepper noise are added in this method. Compared with well-known sobel algorithm and canny algorithm edge detectors in MATLAB, this approach not only detects real boundary of image, but also effectively suppresses the impact of two types of noises on edge detection. This new approach makes a great progress.

Key words: nonparametric change-point statistic analysis, edge detection, computer vision