An Effective Preprocessor of Edge Detection for Distorted Images

摘要

Edge detection is an intuitive strategy to figure out an object in digital images. Nevertheless, if an image is spread with noise such as ultrasonic noise, scanner noise, or camera noise, general edge detectors cannot perform effectively. In this paper, a preprocessor was provided to handle this issue. For each pixel in a speckle image, the preprocessor first determined whether it was noise or not. It then checked if it was located on an edge. According to the information, the preprocessor can define different weighted coefficients to equalize each pixel. After that, we can perform an edge detection mechanism on the processed image to draw the edge map. Experimental results showed that the preprocessor is able to effectively improve the performance of edge detection mechanism in speckle images.