Fast Decision Based Weighted Fuzzy Mean Filtering For Extremely Corrupted Images

Abstract This paper presents an uncertainty based novel filter that integrates randomness and fuzziness for extremely corrupted images by salt and pepper impulse noise. In real time work high computational efficiency is the most important factor while suppressing noise with better edge and detail preservation. The proposed algorithm is bounded with all these aspects. In this scheme the corrupted pixel is identified by the strong decision and replaced it by a weighted fuzzy mean estimation. Since the Certainty Degrees CD of each pixels are soft values and are used as the weights the noisy pixels are reconstructed with an appropriate pixel values. The simulation results show that the performance of the proposed filter is much better than decision based algorithms and cloud model based detection filter across a wide range of impulse noise as high as 90 with edge and detail preservation.