Color edge detection for noisy images by nonlinear prefiltering and block-by-block rotations
This paper addresses a method to obtain color edge detection for images corrupted with Gaussian noise and impulse noise, to correctly reproduce distinct, continuous edges based on nonlinear prefiltering followed by block-by-block rotations to locate the edges in all orientations. A nonlinear prefilter is used to reduce the noise in Red, Green, and Blue components of the color image. The method preserves edges, corners and fine image details, smoothes the noise. Then applied rotation operations block-by-block by convoluting the prefiltered image with 3 × 3 kernel to obtain the edge pixels. The algorithm has tested on a variety of standard images and the performance has been compared with algorithms known from the literature in terms Figure of Merit (FOM).
