模式识别系列讲座

Lecture Series in Pattern Recognition

题    目（TITLE）：Image Rearrangement & Video Synopsis
讲 座 人（SPEAKER）：Prof. Shmuel Peleg, The Hebrew University of Jerusalem,Israel
主 持 人 (CHAIR)：Dr. Veronique Prinet
时    间 (TIME)：14:30PM; Aug. 29 (Monday)
地    点 (VENUE)：The Second Meeting Room, 13th Floor

报告摘要（ABSTRACT）：

Two topics involving image and video rearrangement will be covered, with applications for image editing and for video summarization.

(a) Geometric rearrangement of images includes operations such as image retargeting, object removal, or object rearrangement. Each such operation can be characterized by a shift-map: the relative shift of every pixel in the output image from its source in an input image. We describe a new representation of these operations as an optimal graph labeling, where the shift-map represents the selected label for each output pixel. This graph labeling problem can be solved using graph cuts. Since the optimization is global and discrete, it outperforms state of the art methods in most cases. Efficient hierarchical solutions for graph-cuts are presented, and operations on 1M images can take only a few seconds. Shift-map can also be used for composition of several images.

(b) When we present a video as a 3D space-time volume, we can achieve video summarization by shifting objects in time. Video is summarized because objects that originally appeared at different times will appear together,creating a shorter but more condensed video synopsis. Assuming that the video is captured by a stationary camera as done by security cameras, the summarization process is done in two staged: (i) Moving objects are identified and extracted. (ii) A synopsis video is constructed from the stationary background and the extracted objects, after the objects were shifted in time to occupy a much smaller temporal period. On average one hour of security video is condensed into one minute.

报告人简介（BIOGRAPHY）：

Shmuel Peleg received his BSc in mathematics on 1976 from the Hebrew University, and his PhD in computer science from University of Maryland on 1979. He is with the Hebrew University since 1981, where he is now a professor of Computer science. His research area includes image processing and computer vision. In addition to hundreds of published papers, Prof. Peleg has 15 issued US patents. Technologies developed by Prof. Peleg are used by several companies.