METHOD FOR FEATURE SET SELECTION, METHOD FOR GENERATING VIDEO IMAGE CLASS STASTIC MODEL, METHOD FOR CLASSIFYING AND SEGMENTING VIDEO FRAME, METHOD FOR DETERMINING SIMILARITY OF VIDEO FRAME, COMPUTER-READABLE MEDIUM, AND COMPUTER SYSTEM

摘要

PROBLEM TO BE SOLVED: To automatically classify video according to the contents by calculating a conversion matrix having conversion coefficients to which coefficient positions are related and selecting the coefficient position of a feature set including an entry according to a conversion coefficient in the conversion matrix. ;SOLUTION: A video file 201 is decoded and converted into a rectangular matrix of pixels, and simplified and converted into a smaller-size rectangular matrix of a low-order image to generate the matrix of conversion coefficients, and video features 208 are selected as the conversion coefficient at a coefficient position in the conversion matrix. A classifier 206 receives and inputs respective video features 208 to image class integrating models 202 to 205, and respective frames of the video file 201 are classified into one of image classes represented by image class statistic models 202 to 205. The corresponding image class determined by the classifier 206 is indexes to labeled video 207 of the class.;COPYRIGHT: (C)2000,JPO