Decision based expanding window unsymmetrical trimmed median filter using last processed pixels high density salt & pepper noise removal using median filter using multiple scanning

A new technique of image filtering, for the images corrupted by high density salt & pepper noise, is presented in this paper. Proposed algorithm makes a decision about the noisy or noise free pixel for its further processing. Then it selects a window and chooses the noise free neighbors from this window for median calculation. And if sufficient numbers of noise free pixels ar not found it expands the window for further processing. And if in the first phase of the algorithm it does not find the sufficient number of pixels for median calculation then it does not process it and in the next phase it uses the concept of last processed pixels for the median calculation. The algorithm is tested for many standard and non standard images and found its performance better than the existing techniques of filtering in terms of PSNR, SSIM and IQI.