Shot Boundary Detection Based On Multilevel Color Histogram Differences

With the development of the Internet and video technology, the number of digital video appears explosive growth. It has become a hot research how to analyze the video so as to obtain meaningful information accurately and rapidly from such a large amount of video data. Due to the complexity of video data semantic richness and characteristics, we must analysis of the hierarchical structure of video data to simplify its characteristics and feature extraction, so that on the video analysis possible. Video content structure can be divided into four levels, the video, the scene, the lens, the frame. The shot level is considered the most appropriate level of the video analysis. Video shot segmentation can be used in a variety of areas of video analysis, such as the classification of video, video comparison, content-based video retrieval, etc.Video shot segmentation is one of the most important steps for video data to do a series of analysis, the current shot segmentation method is mainly divided into two kinds: abrupt change, gradual change. Although detection of two lenses are made of good, but in order to achieve fast and efficient shot segmentation, it still need a more in-depth research, especially for gradual shot segmentation research, there are still some problems. The performance in the detection of gradual change is not comprehensive enough, or is often required to extract the features of more complex or repeated testing, resulting in low detection efficiency.Under such a background, this paper comb in recent years based on content-based video retrieval research work based, from video analysis processing flow angle of video shot segmentation are studied and proposed gradual positioning a using multi-level color histogram difference complete gradient lens conversion method for abrupt shot, amplified by gradual transition frames between the difference, then we can easily locate the start frame and end frame and remove in the middle of the gradual transition frames to complete gradual conversion to an abrupt shot. On the basis of the gradual change shot, a three stage method is proposed to detect the shot boundary. Through the multi-stage progressive detection process, in each stage of the shot boundary detection filter the wrong detections, so as to improve the accuracy of the detection results. In addition, by using the three stage detection results, the adaptive regulation of the detection threshold used in the detection process is used to improve the general performance of different types of video shot segmentation. Experimental results show that the method of gradient transformation and multistage detection method has good effect, and it has a strong practical significance for the segmentation of the lens.