视频对象分割及跟踪方法研究

基金项目: 武汉大学科研启动基金资助项目(502273010)

详细信息

Video Objects Segmentation and Tracking in Video Sequences

Key Laboratory of Multimedia and Network Communication Engineering, Hubei Province, Wuhan University, 129 Luoyu Road, Wuhan 430079, China

摘要: 讨论和分析了视频对象全自动提取及后续帧中的跟踪过程及采用的方法,对其算法、分割性能和结果进行了比较和评述。并结合Hausdorff和Snake跟踪器,对初始对象轮廓进行跟踪。结果证明,这种跟踪方法能有效地解决对象被遮挡和形变以及背景移动的问题。
- 视频对象 /
- 对象分割 /
- 对象跟踪 /
- 光流 /
- 活动轮廓 /
- 聚类分析
Abstract: The ISO MPEG4 has attracted much attention recently for providing a standard solution for object-based coding and multimedia data access and manipulation.So content-based representation and coding of the visual information is currently becoming an extremely active research field.Object-based video coding can provide greater compression ratio and better quality of reconstructed images.Video objects segmentation is a key technology in object-based video coding and multimedia data access etc. This paper depictes and analyzes and compares the video objects automatic segmentation and tracking processes and methods.Existing problems and development prospect in this field are described.
- video object /
- object segment /
- object tracking /
- optical flow /
- snake model /
- clustering analysis