摘要 |
<P>PROBLEM TO BE SOLVED: To discriminate shadow areas and overlap parts thereof when a plurality of shadow areas in a moving picture is overlapped. <P>SOLUTION: A shadow area extraction means 102 extracts shadow areas in frames or field of the moving image. An area model is constructed by a model construction means 103 with respect to each shadow area, and the constructed area model is stored by a model storage means 104. Constructed area models are updated per frame or field by a model updating means 105, and overlap of respective area models is discriminated by a model overlap discrimination means 106, and respective areas and overlap parts of them are specified. <P>COPYRIGHT: (C)2008,JPO&INPIT |