Abstract
In computer image analysis (scene analysis), detection of clear edges is a matter of great significance. In this paper, we proposed algorithm of dynamic edge detection, based on discriminant criteria. This program has two steps. At the first step, primitive edges are detected from local area of image. At the second step, final edges are detected in the whole image from the global poit of view. Application of this algorithm to 3 types of images lead the following result. (1) Clear and thin edges are detected from any type of image. (2) Noise edges from texture area are surppressed.
