New Sparsity Models for Image Smoothing and Segmentation

When: Friday, December 11, 2015
Where: PGH 232
Time: 11:00 AM

Speaker: Yizhou Yu, Department of Computer Science, The University of Hong Kong

Host: Prof. Zhigang Deng

Identifying sparse salient structures from dense pixels is a longstanding problem in visual computing. Solutions to this problem can benefit both image manipulation and understanding. In this talk, I first introduce an image transform based on L1-regularization for piecewise image flattening. This transform can effectively preserve and sharpen salient edges and contours while eliminating insignificant details, producing a nearly piecewise constant image with sparse structures. This image transform has been successfully applied to both edge-preserving smoothing and complex scene-level intrinsic image decomposition. Extensive testing on the Intrinsic-Images-in-the-Wild database indicates our method can perform significantly better than existing techniques both visually and numerically.

I further present a new nonlinear embedding, called piecewise flat embedding, for image segmentation. Piecewise flat embedding attempts to identify segment boundaries while significantly suppressing variations within segments. We adopt an L1-regularized energy term in the formulation to promote sparse solutions. We further devise an effective two-stage numerical algorithm based on Bregman iterations to solve the proposed embedding. Piecewise flat embedding can be easily integrated into existing image segmentation frameworks. Experiments on BSDS500 indicate that segmentation algorithms incorporating this embedding can achieve significantly improved results.

This is joint work with Sai Bi, Chaowei Fang, Xiaoguang Han, and Zicheng Liao.

Bio:

Yizhou Yu received the PhD degree from the University of California at Berkeley in 2000. He is currently a full professor in the Department of Computer Science, The University of Hong Kong, and an adjunct professor at the University of Illinois, Urbana-Champaign. He received the 2002 US National Science Foundation CAREER Award, 1998 Microsoft Fellowship and the Best Paper Awards at 2005 and 2011 ACM SIGGRAPH/EG Symposium on Computer Animation. Prof Yu has served on the editorial board of IEEE Transactions on Visualization and Computer Graphics, Computer Graphics Forum, The Visual Computer and International Journal of Software and Informatics. He was the program co-chair of Pacific Graphics 2009, Computer Animation and Social Agents 2012, and has served on the program committee of many leading international conferences, including SIGGRAPH, SIGGRAPH Asia, and International Conference on Computer Vision. His current research interests include data-driven methods for visual computing, digital geometry processing, video analytics, and biomedical data analysis.