Design of Video Retrieval System Using MPEG-7 Descriptors

A Survey on Content Based Video Retrieval Using Mpeg-7 Visual Descriptors

“…A video is made up of number of shots each with boundary properties like cut, fade, dissolve etc [1]. A shot is consecutive frames from the start to end of the recording of video showing continuous action sequence.…”

“…For shot detection, first the features of each frame are extracted, then similarities between the frames using extracted features are measured and finally the shot boundaries are detected between the frames which are not similar. Bin Liang, Wenbing Xiao and Xiang Liu [1] have proposed a CBVR system that consists of three parts: shot boundary detection, feature extraction and similarity measurement. The cut and dissolve have been detected using the histogram difference and skipping image difference, respectively.…”

“…Motion features can easily handle the object and camera motion. Bin Liang, Wenbing Xiao and Xiang Liu have used the colour structure descriptor and edge histogram descriptor for feature extraction for detecting shot boundaries [1].…”

“…Pair wise similarity computes the distance between consecutive frames while window wise similarity computes the distance between frames within a window. In [1] dynamic weighted feature similarity calculation based on the success rate of the visual similarities of different features is used.…”

“…This descriptor is usually combined with the homogeneous texture descriptor [18], to provide a scalable solution to represent homogeneous texture regions in images. The edge histogram descriptor [1] [19] is used to represent the spatial distribution of five types of edges (four directional edges and one non-directional). This descriptor generates the bins of local histograms of these edge directions, which may be combined as global or semi-global histograms.…”