원문정보
초록
영어
Movement information on objects in videos can be used to characterize the content of a scene. This paper proposes a novel shape-based movement-searching algorithm that can effectively search human movements in video streams. Movement information is represented a sequence of boundary shapes of human subjects extracted from input video frames. Information on individual shapes in the sequence is converted into a 1D shape feature by using a shape descriptor. Human movements can be identified in a sequence of shape features extracted from the video, as in the case of searching for words in a long block of text. A comparison of the performance of the proposed algorithm with that of other methods shows that the proposed method can effectively describe human movement information and be useful for movement retrieval applications.
목차
1. Introduction
2. Previous studies of movements
2.1. Temporal templates
2.2. Shape variation descriptor
2.3. Shape sequence descriptor
3. Movement Searching in a Video Stream
3.1. Shape variation search
3.2. Shape sequence search
4. Experimental results
5. Conclusions
References