一种采用分裂Bregman迭代的自适应非本地全变差盲反卷积.pdfVIP

2012 Fifth International Symposium on Computational Intelligence and Design An Adaptive Non-Local Total Variation Blind Deconvolution employing Split Bregman Iteration Huijie Gao Zhiyong Zuo* National Key Laboratory of Science and Technology on National Key Laboratory of Science and Technology on Multispectral Information Processing Multispectral Information Processing Huazhong University of Science and Technology Huazhong University of Science and Technology Wuhan, China, 430074 Wuhan, China, 430074 gaohuijiewhdx@163.com *Corresponding author: zzy.iprai@ Abstract—Total variation has been used as a popular and channel images. However, there is only a single image in effective image prior model in regularization-based image many practical situations. Therefore, blind deconvolution blind restoration, because of its ability to preserve edges. methods based on a single image are worth researching. However, as the total variation model favors a piecewise It is well known that the blind deconvolution is ill- constant solution, the processing results in the flat regions of posed. To avoid such problem, some regularization terms the image being poor, and it cannot automatically balance the should be used. You and Kaveh[2] proposed a regularization processing strength between different sp