基于改进分块颜色特征和二次提取的关键帧提取算法

关键帧提取技术是视频摘要、检索、浏览和理解中的一项重要技术。目前关键帧提取算法存在一些问题,例如特征选择复杂、阈值选择难、自适应性不强等。为了更有效地提取视频关键帧,提出了一种基于改进分块颜色特征和二次提取的关键帧提取算法。首先,对视频帧进行等面积矩形环划分；其次,提取矩形环的HSV量化颜色特征,并由帧图像中心到外依次减小每个矩形环特征的权值以突出图像主体部分；然后,依据相邻视频帧间特征的显著性变化初步选取关键帧；最后,依据初次提取的关键帧在视频中的位置间隔大小进行二次提取优化关键帧。实验结果表明,该方法具有良好的适应性,同时能够有效避免因镜头有突然闪光或物体快速运动而提取过多的关键帧,最终提取的关键帧能够比较全面准确地表达视频内容。

Key Frame Extraction Algorithm Based on Improved Block Color Features and Second Extraction

Key frame extraction is an important technique in video summarization,searching,browsing and understanding.Nowadays some problems exist in the algorithms of key frame extraction,such as selecting features problem,choosing threshold difficultly,week adaptability and so on.In order to extract key frame efficiently,this paper proposed an improved key frame extraction algorithm based on low-level features.Firstly,each frame is divided into equal-area rectangular ring.Secondly,sub-block accumulative color histogram is extracted as color features and different weights are set for different rectanglular rings in order to highlight the central part of frame.Thirdly,key frames are selected in accordance with the significant change of frames.Lastly,key frames are optimized and selected in accordance with the frames of location in the video.The experiment results show that the proposed algorithm has good adaptability and can effectively reduce redundant key frames when the shot has a sudden flash or the object moves fast.Finally,the extracted key frames by this algorithm can express the primary content of video effectively.