Image Salt-Pepper Noise Elimination by Detecting Edges and Isolated Noise Points

摘要

It deals an algorithm for removing the impulse noise, which is also called salt-pepper noise, in this paper. By evaluating the absolute differences of intensity between each point and its neighbors, one can detect the edges, the isolated noise points and blocks. It needs to set up a set of simple rules to determine the corrupted pixels in a corrupted image. By successfully identifying the corrupted and uncorrupted pixels, especially for the pixels nearing the edges of a given image, one can eliminate random-valued impulse noise while preserving the detail of the image and its information of the edges. It shows, in the testing experiments, that it has a better performance for the algorithm than the other's mentioned in the literatures.