基于 JMVC 的立体视频场景突变检测算法

摘要

By using the method of combined judgment of motion vector of the single view and the disparity vector of the stereoscopic video, we proposed abrupt change detection algorithm for stereoscopic video scene.Firstly,the algorithm extracted directly the motion vectors and the disparity vectors from multi-view video coding encoder.Then the average amplitude values of the disparity vectors and the motion vectors of each frame were calculated respectively.Finally,the average amplitude values of the disparity vectors and the motion vectors of each frame were compared with their corresponding threshold values,which could detect the abrupt change of the stereoscopic video scene.The proposed algorithm directly determines the abrupt change of stereoscopic video scene in the compressed domain,therefore,it can not only guarantee detection accuracy,but also reduce computational complexity.The simulation results show that the algorithm is effective.%采用单视图运动矢量与立体视频中视差矢量联合判断的方法，提出一种立体视频场景突变检测算法。该算法首先直接从多视点编码码流中提取视差矢量与运动矢量，然后分别统计每帧的视差矢量与运动矢量的幅度平均值，最后将每帧的视差矢量与运动矢量的幅度平均值与对应阈值相比较，检测立体视频场景突变的情况。用该算法在压缩域直接对立体视频场景突变进行判断，不仅可以保证检测精度，还降低了计算复杂度。仿真结果证明了算法的有效性。