Multispectral Image Denoising Using Curvelet Transform and Kriging Interpolation Based Wiener Filter
The transmission of visual information in the form of images is becoming a major criterion of communication in recent times but the images which are obtained during data acquisition or transmission are seldom bombarded with noise which degrades image quality. This paper uses a curvelet approach for denoising of images using wiener filtering with kriging interpolation. In this the weights of the wiener filter are estimated with the help of the kriging interpolation method in order to get best estimate for removing the noise of the image in a step by step process. Firstly, the corrupted pixels are processed by a clustering mechanism using global patch mixture model of Gaussian called global patch clustering to distinct noisy and non-noisy pixel patches and then an effective transform called curvelet is applied to the patches. The weights are produced by evaluating the semi variance between noisy and clear patches. Finally, the performance of the filter is tested with a noisy image and a relative analysis is performed with existing approaches to prove the robustness of the proposed approach.