MPEG视频码流中I帧快速搜索算法

Fast I-frames search algorithm for MPEG video streams

为解决现有数字视频检索算法中存在的效率不高,在实际应用中存在局限性等问题,提出了一种新型的I帧快速匹配搜索算法。该算法充分研究了MPEG(moving picturesexperts group)码流的结构特点和统计特性,提高了I帧搜索算法定位的准确性;并且使用自适应滤波算法对视频序列I帧的位置进行预测,加速了I帧匹配搜索的过程,提高了I帧搜索算法的效率。仿真实验证明,该算法在效率和视频检测定位准确性方面都优于现有方法;此外,该算法既能保证质量,且计算复杂度不高,经编程实现,具有较强的实用价值。

Existing methods for video indexing are not very efficient or accurate. This paper describes a fast I-frame search algorithm for MPEG （moving pictures experts group） video streams based on the structure and statistical characteristics of the MPEG video stream to achieve better I-frame locating. The algorithm also uses the Least Mean Squares （LMS） method to accelerate the frame search process. Simulations show that the algorithm has higher accuracy than existing methods and ensures quality with less calculations. The effect of this search algorithm is obvious.