Video Shot Boundary Detection and Scene Segmentation
Hong Lu (Fudan University, China), Zhenyan Li (Nanyang Technological University, Singapore), Yap-Peng Tan (Nanyang Technological University, Singapore) and Xiangyang Xue (Fudan University, China)
Copyright: © 2006
This chapter presents a new and efficient method for shot boundary detection (SBD) and scene segmentation. Commonly the first step in content-based video analysis, SBD partitions video data into fundamental units of shots. Over the past decade, SBD has attracted a considerable amount of research attention. However, the detection accuracy achieved leaves much to be desired. In this chapter, a new SBD method based on sequential change detection is proposed to achieve improved detection accuracy. The method is then extended to segment videos into scenes. Compared with existing scene segmentation methods, the proposed method can also obtain more accurate results over a large set of test videos.