Speaker: 刘文（武汉理工大学）

Title: Nonsmooth Regularization and its Imaging Applications

Location：恩明楼数学中心813

Time: 2018.12.21 5:00-6:00pm

Abstract: Image reconstruction usually presnts and ill-posed inverse problem in imaging applications, e.g., deblurring, dehazing, and parallel MRI, ets. To enhance imaging quality, regularization techniques have been proposed to stabilize the image reconstruction process. In the literature, the popular regularized variational model are ofen nonsmooth and easily make the resulting nonsmooth optimization problems difficult to directly handle through traditional numerical methods. According to the specific imaging properties, this talk will present several nonsmooth regularized variational models in different imaging problems. Powerful numerical methods will also be introduced to effectively handle the resulting optimization problems. Experimental results on deblurring, dehazing and parallel MRI have demonstrated the effectiveness of the nonsmooth regularized variational models.