FLAD-Feature Based Locally Adaptive Diffusion Based Image Denoising

A novel patch based adaptive diffusion method is presented for image denoising. This is done with the purpose of locally and feature adaptive diffusion and for attaining patch-wise best peak signal to noise ratio. Our framework uses over-segmentation method to segment the image in to sensible regions and then diffusion of each segment/region to obtain the near-optimal solution and iterates to a lessersegmented region/patches until a best PSNR value is attained. In performing diffusion the method uses the inverse difference moment (IDM) which is a robust feature in determining the amount of local intensity variation in the presence of noise. The experiments show that the proposed method delivers high denoising performance, both in terms of objective metric and the visual quality.