Restoration of Correct Pixel Value via Proposed Adaptive Filter

This paper proposes a new repetitive based Non-linear edge preserving filter which restores the image corrupted by speckle noise in two steps. First a speckle detection algorithm is used before filtering, which separates out the corrupted pixels, and hence only a certain percentage of the pixels will be filtered. Secondly, both speckle detection and noise filtering schemes are applied in progressive fashion through several iterations. Results obtained both by numerical measures such as MAE, MSE, PSNR, ISNR, IEF and manual observation shows that the proposed filter has an appreciably high Peak Signal to Noise Ratio (PSNR) value and at the same time preserves the edges and fine information in images. This method works well for highly corrupted images.