Performance Improvement of Blind Image Restoration Using Ringing Removal Processing
Masahiro Goto, Tomio Goto · 2019
Blur is a representative feature of image degradation. The restoration of a degraded image can be successfully achieved using a method in which point spread function estimation and ideal image estimation processing are alternately repeated. In this paper, we introduce a ringing removal method using L0regularization and propose a highperformance restoration method that can estimate sharper images. It is known that the prior probability of the gradient is effective in suppressing deterioration such as ringing. Thus, ideal image estimation is performed only using gradient information. In the proposed method, a bilateral filter is used to compare the estimated ideal image obtained in previous studies with that obtained using gradient information. By subtracting the difference map from the final deconvolution result, a clear image without ringing could be obtained.