摘要： 根据目前的技术水平，在建立多层分段模型的基础上，识别出视频流中的镜头，利用光流场图象序列分析技术处理镜头中的代表帧及其邻接帧，获取视频数据流中镜头的运动特征。所提出的方法既能获取镜头的平移运动特征，也能获取镜头的旋转运动特征。
关键词:
视频,
镜头,
代表帧,
光流场,
运动特征
Abstract: In accordance with the current technologies and based on multi-level segment model, we recognize the shots in video siream,and use image sequences analyse technologies process r-frames and adjacent frames to acquire motion features of shot in video stream. We can acquire not only transfer motion features but also rotation motion features.
中图分类号:
郑鹏,林晓明,伏玉琛,周洞汝. 获取视频特征的一种方法[J]. , 1999, 25(06): 30-32.
Zheng Peng, Lin Xiaoming, Fu Yuchen, Zhou Dongru. A Kind of Method to Acquire the Features of Video[J]. , 1999, 25(06): 30-32.