Parametric Neural Operator for Non-Line-of-Sight Imaging

发布时间:2025-08-04 供稿单位:数学与统计学院 点击次数:

标题:Parametric Neural Operator for Non-Line-of-Sight Imaging

报告时间:202581日(星期16:00-17:00

报告地点:人民大街校区数学与统计学院415教室

主讲人:段玉萍

主办单位:数学与统计学院

报告内容简介:

Non-line-of-sight (NLOS) imaging is an advanced computational imaging technology aimed at reconstructing obscured or hidden scenes using indirect light signals. These signals are typically generated through multiple reflections or scattering, resulting in weak signal strength and susceptibility to noise interference. Therefore, incorporating physical processes into the reconstruction is crucial for enhancing the quality. We propose a parametric neural operator model capable of learning complex mapping relationships. Through training, this model can simulate the propagation of light and extract useful information from indirect light signals. By leveraging the powerful fitting capabilities of neural networks, this approach can handle complex light transmission models and effectively reduce noise.

主讲人简介:

段玉萍,2012年在新加坡南洋理工大学取得计算数学博士学位,2012年-2015年在新加坡科技研究资讯通信研究院担任研究科学家。2016年任天津大学应用数学中心教授,博士生导师,2023年加入北京师范大学数学科学学院。主要研究方向包括变分图像处理方法和模型驱动的深度学习方法,已在IEEE TPAMI、TIP、TMI、IP、JSC等期刊和CVPR、MICCAI等会议发表论文70余篇,已授权国际/国内专利5项。