Blur - Map Estimation and Deblurring of Degraded Image by Non - Convex Regularization Approach
Latha H N and Rajiv R Sahay · International Journal for Modern Trends in Science and Technology · 2020
We in this work proposed a new algorithm for blur-map estimation of space variant degraded image using a graduated non convex discontinuity adaptive Markov random field (DAMRF) frame work.First we estimated to find the blur-map and then restore the blurred image in.We use a GNC optimization based framework, wherein we assume the latent image and blur map as DAMRF prior.We show that the estimated blur-map is utilized for image restoration application such as image in-painting, depth estimation, image denoising, super resolution, haze removal etc. Robust estimation of the non-uniform blur-map of a natural scene is obtained produces better qualitative and quantitative results compared to the few state-of-the method