Improved TV Image Denoising over Inverse Gradient

Noise in an image can affect one’s extraction of information, therefore, denoising is important pre-processing process. Many the existing models have a large number estimated parameters, which increases time complexity model solution and achieved effect less than ideal. As result, this paper, improved image-denoising algorithm proposed based on TV model, effectively solves above problems. The L1 regularization term make generated by sparser, thus facilitating recovery high-quality images. Reducing while using inverse gradient to estimate enables parameters achieve global adaption improves combination with term. split Bregman iteration method used decouple into several related subproblems, solutions coordinated subproblems are derived as optimal solutions. It also shown that converges Karush–Kuhn–Tucker point. Experimental results show paper more effective both preserving texture structure suppressing noise.