Multispectral Image Denoising With Kriging Interpolation Based Wiener Filter

Abstract

Recently, the application of interpolation to image denoising has shown remarkable denoising results. The multispectral images are mostly corrupted by noise during data acquisition. Therefore, the resulting image is of poor quality. Hence, to improve the image quality, a Kriging Interpolation based Wiener Filter (KIWF) is being introduced in the paper. Wiener filter takes advantage of the kriging interpolation algorithm to obtain the best possible estimate which is obtained by denoising the image. Global patch clustering using Gaussian mixture model is used to separate the noise pixels from clear pixels. The Obtained weight values are applied by estimating the semi variance between the clear patches and noisy patches. Interpolation always helps to suppress very high noise density and achieves good noise reduction. At the end, filter performance is tested and a comparative analysis takes place with the existing denoising techniques to show effectiveness of the filter.