Denoising noise is a basic content in image pre-processing and denoising salt-pepper noise is an important part of it. In view of denoising salt-pepper noise, a weighted mean filter based on decomposition of histograms is proposed in this thesis.Firstly, the behavior of histograms of images under different rates of salt-pepper noises is illustrated, and the relation between histograms of the original image and its noisy versions is analyzed so as to deduce an estimated histogram for the original image from noisy ones. Next, the estimated histogram is decomposed so as to get Gaussian curves nearest to the shape of the histogram. Based on decomposition of histograms, the algorithm of the filter is proposed and is implemented. The experimental results suggest that the filter is simple and effective. |