Speaker: Professor Wang Hanzi (Xiamen University)
Time: At 9:00 on October 18, 2010
Venue: 3rd Floor, Building 3, Optical image Analysis and Learning Center (OPTIMAL) Conference Room

Abstract:
Robust statistical methods is vital when solving problems in computer vision. It is necessary to realize that the data from image or image sequence may be inaccurate when it comes to practice. Data will almost inevitably be polluted by outliner. Outliner is possible consequence of noise of the sensor, data errors, feature extraction failure, feature matching errors, segmentation errors and other factors. Moreover, the data may contain multiple structures. Thus, in the field of computer vision, many have acknowledged that it is a must to apply robust statistical method for all computer vision algorithms as for accurate estimation reason. But it is still an important yet challenging task to construct an accurate parameters structure that contains a large number of external outliner with multiple structures. This report introduces the work of the speakers in terms of robust statistics application, and some applications of computer vision: including motion estimation and motion segmentation, range image segmentation, fundamental matrix estimation, camera pose estimation, medical image processing, reconstruction, background / foreground appearance modeling, video tracking and video segmentation.

Speaker Profile:
Wang Hanzi, male, born in 1973, Xiamen University, Minjiang Scholar/Professor. In 1996 and 1999, respectively, he graduated with his bachelor's and master's degree from Sichuan University. In 2004, he graduated from Monash University, Australia with his PhD, and won the honor of Douglas Lampard the Best Dissertation Award. From 2004 to 2006, he worked in Australia Monash University, computer systems engineering department as a research fellow. From 2006 to 2008, he worked in the Johns Hopkins University USA, Computer sciences department as a postdoctoral fellow and assistant research scientist respectively. From 2008 to2010, he worked in School of Computer Science, University of Adelaide in Australia, as a Senior Research Fellow. In 2010 he was selected as a Min jiang Scholar Professor in Xiamen University. His Research areas including artificial intelligence, computer vision and pattern recognition, image and video processing, video tracking and monitoring, robust statistics and model fitting, object detection, object recognition, motion estimation and segmentation, optical flow computation, multi-view geometry science, three-dimensional reconstruction and so on. His nearly 50 research papers have been published in both domestic and foreign academic journals and international conferences. Three of his research papers have been published in international journals IEEE Trans. PAMI and two of his papers have been published in IJCV. There are also many of his articles published in international authoritative journals IEEE Trans, CSVT, PR. as well as top international conference such as ICCV, ECCV, CVPR, NIPS, MICCAI. He is also the associate editor for IEEE Trans. CSVT.

Optical image analysis and Learning Center
Tel: 02988889302
E-mail: OPTIMAL@opt.ac.cn