MPEG压缩域关键帧提取的研究  

Keyframe Extraction in MPEG Compression Domain

作　　者：邱多[1] 王成儒[1] 

机构地区：[1]燕山大学信息科学与工程学院,河北秦皇岛066004

出　　处：《计算机安全》2011年第1期14-16,共3页Network & Computer Security

摘　　要：提出了一种基于MPEG压缩域关键帧的提取方法。首先对MPEG压缩域中实现I帧的提取,然后利用I帧的特征8╳8块的DC系数,计算相邻I帧对应位置的8╳8块的DC系数相似性度量,进而实现镜头的判断,最后利用P帧宏块类型提取关键帧。由于处理过程是直接从MPEG的压缩视频提取,不需要对其完全解压,所以计算复杂度低,提高了提取速度。实验证明该方法效率高,较好地代表了视频内容。In this paper,keyframe extraction algorithm based on MPEG compressed domain is proposed.First extract the I-frame from MPEG compressed domain,and calculate the similarity measure of 8＊8 DC coefficient between the adjacent I-frames,and compare the macroblocks of P-frame to extract the keyframe.Our approach is calculated in MPEG compressed domain indirectly,which needs not absolutely to be uncompressed,computational complexity is reduced,and the processing speed can be improved.The result show that our approach is computational simple and able to adapt to the visual content.

关 键 词：压缩域 关键帧 DC系数 宏块类型 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]