ABSTRACT
Edge detection plays an important role in image processing. One of the main goals of the edge detector is to select the appropriate detection thresholds to sharpen the image edges and benefit to the information extraction. Using dierent edge detection operators can get dierent detection eects. The traditional Canny operator edge detection is to deal with the whole image directly, whether using the genetic algorithm to find the double threshold of Canny operator, or using the adaptive method to change the weights of the Gaussian filters, both of them are aimed to find the optimal value for processing objects from the perspective of determining parameters (He 2013, Deng 2013). But for the whole image, the features of every part of the image are dierent, and its gray scales will change in a large range. If we directly use the traditional edge detection method for the whole image, it will miss some edge information. In this paper, a block method for the whole image has been used in the edge detection process, and using the gray level statistic values to select the optimum thresholds have been introduced in the process of edge detection.
