A new method for edge detection in digital images

Edge detection is a very difficult task. In the paper a new method, referred to as the anisotropic Gaussian kernel, is proposed. The algorithm selects every image pixel´s direction derivation for image edge detection. Unlike the existing isotropic-type way that is widely used in image edge detection, the new technique depends on both the anisotropic Gaussian kernel and the image´s direction derivative information. The detection performance of the new algorithm is better than the classical Canny detection algorithm under the same noise environment.