當前位置:

Rediscover Singular Value Decomposition for Image and Video Processing


講座名稱:Rediscover Singular Value Decomposition for Image and Video Processing

講座時間:2014-12-25  15:30

講座地點:主樓2區227

講座人:Weisi LIN

講座人介紹:


Lin Weisi graduated from Sun Yat-Sen University, Guangzhou with B.Sc and M.Sc, and from King’s College, London University, UK with Ph.D. Currently, he is an associate professor in School of Computer Engineering, Nanyang Technological University (NTU) in Singapore. He was the Lab Head of Visual Processing in Institute for Infocomm Research (Singapore) before joining NTU. His areas of expertise include video compression, image quality evaluation, and perceptual signal modelling. He has served as an AE for IEEE Trans. on Multimedia, IEEE Signal Processing Letters and Journal of Visual Communication and Image Representation.  He holds 7 patents, published 100+ journal papers (including 60 IEEE Trans papers) and 190+ conference papers, authored 2 books, edited 3 books, and wrote 9 book chapters; he has been elected as a Distinguished Lecturer of Asia-Pacific Signal and Information Processing Association (APSIPA). He guest-edited 7 special issues in international journals, and organized 15 special sessions in international conferences. He has served as a technical program chair for PCM 12, ICME 13, QoMEX 14 and PV 15, and the ICME Steering Committee (2014-2015).

講座內容:Singular Value Decomposition (SVD) has been known for long, and already used extensively in different signal representation, processing and understanding tasks. We believe that as a general tool for any matrix (to represent any 2D signal) SVD needs to be understood more. Therefore, with our recent research projects, we have re-visited the topics related to SVD, discovered more insights and explored further potentials in image and video manipulations. It is hoped that this talk would trigger more discussion into additional research opportunities along these directions, especially to enable a common framework for various visual signal processing tasks