Variational Methods and Learnable Methods for Image Restoration

报告人： 刘九龙（中国科学院）

时间：2022年5月19日 下午19：30-21：30

地点：腾讯会议888-117-539

摘要：Image restoration is a family of inverse problems for recovering high quality images from corrupted observations. They are fundamental problems in image science, and they also provide a testbed for more general inverse problems. As the advances in imaging modalities, the observations trend to be with low-cost data acquisition and the image restoration problems trend to be high-scale. The image restoration is thereby still a popular topic in image science and computational mathematics. In this talk, I will introduce some traditional variational methods and recent learnable methods for image restoration， and finally I will also present some of our recent work on medical image reconstruction problems, phase retrieval, and some other nonlinear image restoration problems.

报告人简介：刘九龙博士目前是中国科学院数学与系统科学研究院计算数学与科学工程计算研究所的副研究员。 他于2017年获得上海交通大学数学博士学位，2018年至2021年在新加坡国立大学数学系从事博士后研究。主要研究方向包括变分方法、优化算法和机器学习理论以及探索它们在医学图像重建和数据处理中的应用。他在SIAM Journal on Imaging Sciences、Inverse Problems、IEEE Transactions on Medical Imaging 等杂志以及ICLR、 CVPR、 MICCAI等机器学习会议上发表了十多篇相关研究的论文。