High Density Salt and Pepper Noise Removal Using Fusion Model of Nonlinear Filtering Techniques

In this paper, an effective technique for removal of high density salt and pepper noise through decision based unsymmetric partial trimmed adaptive neighborhood mean filter is proposed. The proposed algorithm comprises of two-phase scheme. The first phase involves Decision Based Partial Trimmed Global Mean Filter (DBPTGMF) and Decision Based Unsymmetric Trimmed Winsorized Mean Filter (DBUTWMF) to work in all possible cases and the second phase is employed after the first phase to replace the left over noisy picture elements (pixel). The proposed algorithm is the fusion of the benefits of DBPTGMF, DBUTWMF and Decision Based Adaptive Neighborhood Mean Filter (DBANMF). The proposed algorithm (PA) is examined against several grayscale pictures in terms of Peak Signal to Noise Ratio (PSNR), Image Enhancement Factor (IEF) and Structural Similarity Index Matrix (SSIM) values at high noise density and it has been observed that the P A gives better results than the Adaptive Median Filter (AMF), Decision Based Algorithm (DBA), Modified Decision Based Unsymmetric Trimmed Median Filter (MDBUTMF), Enhanced Modified Decision Based Unsymmetric Trimmed Median Filter (EMDBUTMF), Modified Decision Based Unsymmetric Trimmed Global Mean Filter (MDBUMF_GM).