TRACKING OF SCENE IN VIDEO BY USING JOINT COLOUR AND TEXTURE HISTOGRAM METHOD

Video categorization needs the economical segmentation of video into scenes. Object pursuit is one in every of the key technologies in intelligent video police work and the way to describe the moving target could be a key issue. a unique object pursuit rule is conferred during this paper by victimization the joint color texture bar graph to represent a target then applying it to the mean shift framework. The video is initial segmental into shots and a collection of key-frames is extracted for every shot. Typical scene detection algorithms incorporate time distance in an exceedingly shot similarity metric. within the technique we have a tendency to propose, to beat the problem of getting previous data of the scene period, the shots square measure clustered into teams primarily based solely on their visual similarity and a label is assigned to every shot in line with the cluster that it belongs to. Then, a sequence alignment rule is applied to discover once the pattern of shot labels changes, with the exception of the traditional color bar graph options, the feel options of the article are extracted by victimisation the native binary pattern (LBP) technique to represent the article. the foremost uniform LBP patterns square measure exploited to create a mask for joint color-texture feature choice. Compared with the normal color bar graph primarily based algorithms that use the entire target region for pursuit, the projected rule extracts effectively the sting and corner options within the target region, that characterize higher and represent a lot of robustly the target. The experimental results validate that the projected technique improves greatly the pursuit accuracy and potency with fewer mean shift iterations than customary mean shift pursuit. Experiments on TV-series and flicks additionally indicate that the projected scene detection technique accurately detects most of the scene boundaries whereas protective an honest trade-off between recall and preciseness.