Edge Detection of Image on the Local Feature

Abstract

The traditional methods of image edge detection are sensitive to the noise comparatively and can not locate the edge exactly. Aiming at the shortage of the traditional methods of image edge detection, the paper puts forward the edge detection of the low signal noise ratio image (the low SNR image) according to the local feature. Firstly, we analyze the relationship of the neighborhood of pixels and introduce the function expression which is used to estimate whether the pixel is on the image edge or not. The method takes the pixel as the research object, and the edge to be detected is perhaps the noise or just itself. According to the edge's continuity of the image and the isolation of noise, we use the morphology to extract the image edge and overcome the noise effect on the edge, then use the morphological gradient to eliminate the false edge points. With the new method, the image edge can be detected more correctly.