A wavelet frame coefficient total variational model for image restoration
share
In this paper, we propose a Rudin-Osher-Fatemi(ROF)-like model for image restoration which utilizes total variation (TV) regularization on the wavelet feature images. The model imposes more smoothing power on the cartoon image generated by the low-pass filter and less strength on the edges generated by the high-pass filters. Thus, the model can preserve more edges and details than the ROF model. Next, the existence of solution for the model was proved and a slightly modified split Bregman algorithm was used to solve it. At last, we present some experimental results to show its competitive advantage to the related methods both in quality and efficiency.
READ FULL TEXT