Image Restoration Using CPN and SOM

This work mainly deals with the restoration of an image degraded by the Impulsive noise, Speckle noise and Gaussian noise. Two types of networks are used to restore the image they are Counter Propagation Networks (CPN) and Self-Organizing Maps (SOM). The parameters in the weight controller are adjustable using learning algorithm. The weights are then substituted to the Center Weighted Median filter for restoration. In SOM Wiener Filter is used to smooth the image degraded by any of the noise mentioned above. The smoothened image is passed to SOM for restoration. Additionally the SOM was applied to the Texture images to restore the poor quality of the image. Experimental results are shown in this paper