The Detection And Extraction Of Moving Object In Video

In the past decade, the technology of video object detection andextraction has been widespread concerned in the field of computer vision.Video coding method based on the content and utilization of human visualcharacteristics of the2nd generation coding technique has been proposed.Video object extraction can greatly improve the compression efficiency, andprovides a convenient storage and transmission. Object-based retrieval andbrowsing technology has been developed by the MPEG-4standard MPEG-7standard. In the field of web technology of the Internet, we need to extractthe video object to query and interact with the static or dynamic scenes. Inaddition, in the field of pattern recognition, computer vision, video retrievalhas also been widely used.MPEG-4is the new content-based multimedia data compressionstandard. It is the first time to propose the object-oriented video code.However, MPEG-4standard only defines a video coding and decodingprocess, and does not develop specific video object segmentation. As animportant assistive technology in the video processing domain, the researchabout video object segmentation has the profound significance and the greatapplication value.This paper presents an image segmentation algorithm combining edgedetection and watershed segmentation based frame segmentation (spatialsegmentation), on the basis of a detailed analysis of existing videosegmentation algorithms. Firstly, it utilizes edge detection operator forimage edge detection and edge point limit, to avoid the region ofover-segmentation, through improved watershed segmentation to supplymissed the edge, in order to get better segmentation results. Based on theinterframe segmentation (time-domain segmentation), this paper analyzesthe basic methods and theories of the moving objects detection andextraction, and adopts adaptive movement object extraction using acombination of three consecutive frame difference and background imagedifference to improve the efficiency of the algorithm. Experiments showthat this algorithm can accurately and quickly detect and extract movingobjects and has strong robustness and broad applicability, provides a solid theoretical foundation for the video surveillance objects’ compression andinquiry, and has good practical value.