Adaptive Filtering Based on Gradient Information

In this paper, an adaptive smoothing filter algorithm based on gradient information is proposed. The new method solves the problem of conventional filer that can’t smooth noise and sharp edge simultaneously. It is based on the iterative convolution of local adaptive template and the original image signal, the template has the property of diffusing anisotropically. In each iteration, the weight coefficients of the filter are determined by the gradient function of each pixel, and they vary with the variety of the gradient function, thus reflects the degree of continuity of the gray value. The weight coefficients also depend on one parameter, which controls the amplitude of the breaking point that needs to be preserved during the iteration. This algorithm sharps the edge of image by iteratively computation, after several iterations the image is adaptively smoothed according to the edge blocking. The simulation results demonstrate that this algorithm can perform filtering effectively, it has appropriate computation complexity and is suitable for real-time processing. Key-words: Image filtering, gradient Information, adaptive smoothing, smooth noise, sharp edge