上海交通大学学报

上海交通大学学报（自然版） ›› 2011, Vol. 45 ›› Issue (10): 1542-1546.

• 自动化技术、计算机技术 • 上一篇    下一篇

基于双重检测模型的视频镜头分割算法

巢娟1，孙锬锋1,2，蒋兴浩1,2   

  1. (1. 上海交通大学 信息安全工程学院， 上海 200240；2. 上海市信息安全综合管理技术研究重点实验室， 上海 200240)
  • 收稿日期:2010-10-30 出版日期:2011-10-31 发布日期:2011-10-31
  • 基金资助:

    国家自然科学基金项目(60802057，61071153)，上海市青年科技启明星计划资助项目(10QA1403700)

A Video Shot Segmentation Algorithm Based on Dual-detection Model

 CHAO  Juan-1, SUN  Tan-Feng-1, 2 , JIANG  Xing-Hao-1, 2   

  1. (1.School of Information Security Engineering， Shanghai Jiaotong University， Shanghai 200240，China； 2.Shanghai Information Security Management and Technology Research Key Lab，Shanghai 200240，China)
  • Received:2010-10-30 Online:2011-10-31 Published:2011-10-31

摘要： 提出一种基于自适应双重检测模型的视频镜头分割算法.该算法分为初检和复检两阶段，初检提出了基于人类视觉特征的非均匀分块的概念，并结合滑动窗口自适应二分查找算法进行镜头边界检测；复检过程利用尺度不变特征变换算法对初检得到的镜头边界进行匹配，排除初检中的误检.视频测试结果表明，该算法对多种不同类型的视频能同时提高切变和渐变镜头边界检测的效果.

关键词:  , 镜头分割, 双重检测, 非均匀分块, 自适应二分查找, 尺度不变特征变换算法

Abstract: Video shot segmentation is the fundamental work of content-based video retrieval. An adaptive shot segmentation scheme based on the dual-detection model was proposed. In the pre-detection round, the uneven blocking mechanism based on human visual features is presented and applied to adaptive binary search (ABS) method in moving window to detect shot boundaries. In the re-detection round the scale invariant feature transform (SIFT) algorithm will match the boundary frames and exclude false detections. The experiments indicate that this algorithm performs well on both abrupt and gradual shot boundary.

Key words: shot segmentation, dualdetection, uneven blocking, adaptive binary search, scale invariant feature transform (SIFT) algorithm

中图分类号: