Blind Deblurring Based On Locally Weighted Total Variation And Continuous Blur Kernel

With the progress of economic globalization and the rapid advance of science and technology, a large number of portable digital products become essential for people’s daily life. However, the degradation of image quality caused in the process of using these devices is inevitable and an urgent problem affecting people’s lives. Furthermore image blurring is the most common in the image degradation and is hotspot in the field of computer vision and image processing. Recently, a large number of motion deblurring methods have been proposed and improve the level of motion deblurring. But how to get high-quality image restoration is still a challenging problem.Image deblurring is an ill-posed problem that requires regularization to improve the stability of the solving process. Firstly, starting from the Local features of the image, this paper proposes a novel Locally Weighted Total Variation (LWTV) regularization method to solve the non-blind motion deblurring, and an effective algorithm is presented based on Alternating Minimization. Compared with the traditional method of TV, LWTV has better adaptability to image structure, and the constructed energy functional can better describe the sharp image (i.e., in this model the gradient energy of sharp image is lower than the blurred image). Secondly, combined with LWTV model, we propose an original method of kernel estimation based on sparsity and continuity of PSF under the guidance of the significant structure of image. And this paper gives the corresponding iterative algorithms and successfully solve relatively accurate blur kernel with sparsity and continuity. Finally, Based on the above two aspects this paper gives to the method of blind deblurring based on LWTV and continuous blur kernel, this method can relatively rapidly estimate the clear image without the noise and blur.Experiments show that the proposed deblurring method can not only remove the blur and noise, but also keep the sharp edge and suppress ringing artifacts. And the experiments show that this method can well restore the blurred image even for large blurs and is an effective approach to motion blur.