Image Restoration Using Four-directional Total Variation with Overlapping Group Sparsity
Total variation has good performance of preserving image features(edge,texture pattern),but also produces some unwanted staircase artifacts.To solve this problem,this paper proposes a new model of four-direction overlapping combination sparse total variation.First,the gradient information of each pixel in the four directions is combined to form a combined gradient,it fully mines the gradient information of the four directions on the image,further restrains the step effect,completely eliminates noise,elimination,improves the quality of the restored image,then uses the alternating direction multiplier algorithm(ADMM)to split the complex optimization problem into easier-to-solve sub-problems,and uses two-dimensional fast Fourier transform(FFT)and optimization minimization algorithm(MM)to alternate iteratively to solve sub-problems,which improves the speed of image restoration.
image restorationoverlapping group sparsityfour-directional total variationADMM
国家科技期刊平台
NSTL
万方数据
