A Novel Image Denoising Algorithm Based on Non-uniform Rectangular Partition and Interpolation

This paper proposes a novel denoising algorithm according to the image approximation after the Non-Uniform Rectangular Partition. A given image can automatically be partitioned into different sub-rectangles with different dimensions and the bivariate polynomial is used to do the Optimal Quadratic Approximation to gray values of image in each sub-rectangle. When the approximation error and bivariate polynomial are specified, a specific image partition result is obtained. Generally, the smallest the error, the better approximation effect is obtained. However, a suitable error should be selected to get the best approximation to original image instead of the noised image. On the other hand, in order to avoid the square effect after denoising and obtain a better denoising result, the interpolation method is used before and after the denoising by Non-Uniform Rectangular Partition. Experimental results show that this method can obtain a better denoising effect by comparing with other methods.