关键帧定位技术有助于提升微视频资源的利用率,在对微视频资源中的关键帧定位过程中,采用当前算法对关键帧定位时,图像帧中的关键特征过多,无法运用少量特征代表定位特征的唯一性,存在图像帧定位误差大的问题.提出一种基于改进帧间像素灰度差值的微视频资源中的关键帧定位方法.该方法先得到微视频资源图像序列,并获取微视频中相邻帧图像差值信号,利用帧间差值信号描述微视频资源中关键帧图像,将相邻帧间差值信号转化为帧间像素灰度差值,在此基础上融合滑动窗口机制,计算出微视频帧间相关性与其自相关性的相对偏差,对该偏差和给定的关键帧判别函数相比较,利用对比的结果完成微视频资源中的关键帧定位.实验仿真证明,所提算法进行关键帧定位的精度在97％.%The authors of this article propose a location method for key frame in micro-video resources based on difference of pixel gray among modified frames.Image sequence of micro-video resources and acquired image difference signal of neighbor frame was obtained.The difference value of inter-frame pixel gray was converted using signal of inter-frame difference to describe image of key frame.On that basis,we worked out relative deviation between inter-frame correlation of video and its autocorrelation integrated with slide window mechanism,then compared the deviation with given discriminant function of key frame.Finally,we used the result of comparison to complete location of key frame.Experimental simulation shows that the presented method has 97％ precision of key frame location.
展开▼
