主题：图信号处理及应用 Graph Signal Processing and Its Applications
As a novel representation of signals that live on an irregular structure, graph provides new potentials to analyze and understand signals over a network. Graph signal processing (GSP) is dedicated in the development of such tools and their applications. E.g. Via graph spectral analysis, the signals on a graph can be converted from vertex domain to a spectral domain and conduct frequency analysis just like a conventional signal. After briefly introducing the basic concepts in graph signal processing, we will use several applications to demonstrate the potential use of GSP for actual problems, including 3D image denoising, motion segmentation, and point cloud compression.
Dong Tian received the M.Eng. and B.Eng. degrees on automation from the University of Science and Technology of China (USTC) in 1998 and 1995, respectively. He holds the Ph.D. degree at Beijing University of Technology in 2001. He is currently a Senior Principal Research Scientist with the Multimedia Group of Mitsubishi Electric Research Laboratories (MERL) at Cambridge, MA. Prior to joining MERL in 2010, he worked with Thomson Corporate Research at Princeton, NJ since 2005, where he was devoted to H.264/MPEG AVC encoder optimization and 3D video coding/processing, especially to the standards of Multiview Video Coding (MVC) and later on 3D Video Coding (3DV) within MPEG. From Jan 2002 to Dec 2005, he was a postdoc at Tampere University of Technology in Finland funded by Nokia Research Center and had made contributions on video coding standard and its applications for mobile environments. His current research interests include graph signal processing, point cloud processing, machine learning, image/video coding and processing. Besides academic publications, he holds over 20 US-granted patents. He is a senior member of IEEE.