Abstract：In this talk, we will integrate mixture models (statistics) and regularization technique (PDE) to develop several algorithms for image segmentation and restoration. Some new adaptive fidelity terms and nonlocal regularizers will be introduced by analyzing a well-known expectation maximization process. The well-posedness theory for the proposed methods will be discussed. We will give some examples of the applications for the proposed methods.

报告人：刘君，北京师范大学数学学院，副教授，主要从事图像处理和数值计算方面的数学研究，个人主页http://math0.bnu.edu.cn/~liujun/。