H.264 Video And Its Dynamic Target Detection

H.264 video and the problem of moving object detection in it are studied in this paper.Moving object detection in videos is a classical topic in the field of computer vision and is widely used in practice. Because video sequences contain lots of information and huge storage space as well as wide transmission bandwidth is highly demanded, video's storage and transmission are encoded. Among present coding and decoding standards, H.264 is the most advanced, which gives the best compression efficiency and network serviceability. Some characters are introduced in this paper. Moving object detection in compressed H.264 videos has a broad practicability.Some traditional detecting methods based on compressing region and uncompressing region are introduced in this paper. In addition, some traditional methods are tested and analyzed. The motion estimation technology can eliminate the inter frame redundancy in video compressing, in which motion vectors that contain motion information of macro block are used. The basic idea of moving object detection in videos is to use motion vectors.In this paper, also the motion vectors of each frame are handled and background motion compensation is used to check. The extended vector median filter is modified and used to filter the noise in motion vector field. Background motion compensation is adopted for moving backgrounds. Tree split disposal is used for complicated background. Template of four surrounding neighborhood regions is used to clean noise blocks in pictures and improve the consistency of motion vectors. Seven experiments on typical videos demonstrate the effectiveness of the algorithm of this paper.At the end of this paper, the summing-up and insufficiencies of the algorithm are listed. This paper will hopefully provide some useful information for later researchers.