PROJECT TITLE :

Gradient Histogram Estimation and Preservation for Texture Enhanced Image Denoising - 2014

ABSTRACT:

Natural image statistics plays an important role in image denoising, and various natural image priors, including gradient-based mostly, sparse illustration-primarily based, and nonlocal self-similarity-based mostly ones, have been widely studied and exploited for noise removal. In spite of the great success of many denoising algorithms, they have a tendency to smooth the fine scale image textures when removing noise, degrading the image visual quality. To address this downside, during this paper, we propose a texture enhanced image denoising technique by enforcing the gradient histogram of the denoised image to be close to a reference gradient histogram of the initial image. Given the reference gradient histogram, a completely unique gradient histogram preservation (GHP) algorithm is developed to enhance the feel structures while removing noise. Two region-based variants of GHP are proposed for the denoising of pictures consisting of regions with totally different textures. An algorithm is additionally developed to effectively estimate the reference gradient histogram from the noisy observation of the unknown image. Our experimental results demonstrate that the proposed GHP algorithm will well preserve the feel look in the denoised images, making them look additional natural.