Multi-level Edge Detectors Based on Convolution Matrices of Base-lengths 2 and 3

Proposed here is a new approach of edge detectors using edge-detecting masks generated from the basic edge detecting masks and convolution matrices of base lengths 2 and 3. The convolution matrices can be generated from a base unity matrix. These matrices behave like a low pass filter, which are effective in reducing noise in the edge detection process. The size of the convolution matrices increases with the increase in the levels. The higher the level, the generated edge detectors produce more prominent edges and have ability to suppress more noise. Experimental results show the superiority of the proposed scheme than most of the existing edge detection algorithms for different kinds of images. Algorithmically, it is simple and easy to implement.