报告时间：2024年6月7日 （星期五）15：00 – 17：00

报告地点：琶洲实验室主楼 一楼学术报告厅

报告题目：Neural Expectation Maximization for Self-supervised Blind Image Deblurring

报告人：新加坡国立大学 纪辉教授

报告摘要：

When taking a picture, any camera shake during the shutter time can result in a blurred image. Recovering a sharp image from the one blurred by camera shake is a challenging yet important problem. Most existing deep learning methods use supervised learning to train a deep neural network (DNN) on a dataset of many pairs of blurred/latent images. In this talk, we will present a dataset-free deep learning method for removing uniform and non-uniform blur effects from images of static scenes. Based on a DNN-based  re-parametrization of the latent image and blur kernels, a Monte Carlo Expectation Maximization (MCEM) approach is presented to train the DNN without requiring any latent images. The Monte Carlo simulation is implemented via Langevin dynamics. Experiments showed that the proposed method outperforms existing methods significantly in removing motion blur from images of static scenes.

报告人简介：

Dr. Ji Hui received his Ph.D. in Computer Science from the University of Maryland at College Park in 2006. He joined the Department of Mathematics at NUS in the same year  and is currently a Professor at NUS. He also serves as the Director of the Centre for Data Science and Machine Learning (DSML) and the Director of DSML graduate program at NUS. His research interests lie in wavelet theory, computational vision, imaging science, and machine learning.