METHODS AND APPARATUSES FOR DETECTING SCENES IN A VIDEO MEDIUM

摘要

Methods and apparatuses of detecting the beginnings and endings of scenes in video media, especially videotapes of "Reality TV" scenes, are disclosed. The video medium defines an array of pixels, and comprises a sequence of video frames. Each frame has a set of pixel data values for representing an image, with each pixel data value being associated with a pixel. Exemplary methods and apparatuses select a plurality of video frames from the video medium, and obtain the pixel data values of a subset of pixels of each selected frame. The pixel subsets of at least one pair of successive frames have a difference in the selection of pixels. A dispersion signal (106) representative of the dispersion in the obtained pixel data values over a set of sequential frames is generated, and a signal is generated to indicate an end of a scene when the dispersion signal falls below a threshold level (108).