摘要
针对基于内容的视频检索系统,提出了一种关键帧提取算法.为了提高算法的保真度和压缩比,首先构造了动态帧,它的每个像素对应一个像素代表灰度集合,该集合中的元素可以最大限度地代表镜头中相应像素的灰度值,然后根据镜头中每一帧与该镜头动态帧之间的距离来确定关键帧.为了验证本算法的有效性,选取了大量视频与TMOF算法及SKF算法进行比较.结果表明:该算法具有较高的准确性和可靠性,保真度和压缩比均高于其他两种算法.
A key frame extraction algorithm was presented according to content-based video retrieval system. At first,a dynamic frame was constructed to improve the fidelity and compression ratio of the algorithm. Each pixel in dynamic frame corresponds to a representative gray set,which means the elements in the set represent the furthest corresponding pixel gray value in the shot. Then the distance between each frame and the dynamic frame was calculated to determine the key frames. The algorithm was tested on comprehensive videos,and compared with both TMOF algorithm and SKF algorithm. Experiments show that the key frame extraction algorithm is more accurate and reliable,and the fidelity and compression ratio are higher than that of the other two algorithms.
出处
《天津科技大学学报》
CAS
2009年第4期69-72,共4页
Journal of Tianjin University of Science & Technology
基金
天津科技大学科学研究基金资助项目(20080204)
关键词
视频检索
镜头
关键帧提取
动态帧
video retrieval
shot
key frame extraction
dynamic frame
