Using structural similarity index measuring the Patch locally optimal wiener filter in discrete wavelet transform

Abstract

Image denoising has been a well studied quandary in the field of image processing and it is still a conundrum for researches. More expeditious shutter speeds and higher density of image sensors (pixels) result in higher calibers of noise in the captured image, which must then be processed by denoising algorithms to yield an image of acceptable quality. In this paper, we propose a method to denoise the images predicated on Discrete Wavelet Transform and Wavelet Decomposition utilizing PLOW (Patch Predicated Locally Optimal Wiener Filter). Transformation and Decomposition provide the approximation and detailed coefficients, for reconstructed image PLOW technique is applied. The patch-predicated wiener filter exploits the patch redundancy for image denoising. It utilizes photometrically, geometrically and graphically homogeneous patches to estimate the different filter parameters. This describes how these parameters can be accurately estimated directly from the input strepitous image. The denoising framework can additionally be generalized to exploit such photometric redundancies within any given strepitous image. Our noise abstraction system utilizes the LARK features which amend the finer estimates of pixel value and its gradients of pristine image. Experimental results demonstrate that our proposed study achieves good performance with reverence to other denoising algorithms being compared. Experimental results are predicated on Peak Signal to Noise Ratio (PSNR), Mean squared error (MSE) and Structural Homogeneous attribute Index Measure (SSIM).