Nonlinear Filters Based on Support Vector Machines

In this work, a new family of nonlinear filters based on support vector machine is presented. This new filter, called support vector machine filter (SVMF), is based on the general concept of binary filters and machine learning theory. Two applications that show the potential of these filters are designed. As a first application, the proposed filter is used as an impulsive noise image denoising. The second application presents a new edge detection structure using a different point of view from the traditional ones. The results obtained for the applications at hand show that the proposed filter outperforms center weighted median in the image denoising task and the traditional edge detectors. The proposed filter can be successfully applied for the processing of images corrupted with impulsive noise while maintaining the visual quality and a low reconstruction error.