Efficient object detection technique in MPEG videos

Abstract

In this paper we propose an efficient approach for Motion Vector (MV) based object detection in MPEG-2 video streams. Rather than processing the whole MV fields that are directly extracted from MPEG-2 video streams in the compressed domain, we perform MV selection process, those MVs have more robust object information, and are know to be more texture. As a result, the object detection algorithm is more capable of accurately detecting objects and efficiently produces the result. We compare the performance of our proposed system with the popular and commonly used object detection techniques in object detection domain. Based on experimental results performed over the MPEG7 testing dataset and measuring performance using the standard recall and precision metrics for perceptual performance, and run time metric for efficiency performance object detection using our texture filter is remarkably superior to the alternative techniques. In addition to these results, we describe a user system interface that we developed, where users can maintain the parameters interactively.