报告题目：Image Segmentation Based on Shape Knowledge

报告人：信息与计算科学系 曾勋勋

时间：2016.9.9(星期五) 16:00

地点：数计学院4#229

Abstract：

For most real applications of image segmentation, purely low-level information such as intensity, color and texture does not provide the desired segmentations. This naturally leads to a need for integrating low-level features and high-level information. For segmenting familiar objects, it is a common way to appropriately incorporate shape priors of objects to facilitate segmentation. Numerous approaches have been proposed for this problem. However, it remains a difficult challenge due to various disturbing factors such as noises, partial occlusions, missing information, background clutter, etc. This talk introduces related works about object segmentation based on shape knowledge and focuses on the problem of how to exploit circulant shift for shape registration in object segmentation.