METHOD FOR DISCRIMINATING TOKEN IN VIDEO SEQUENCE

摘要

PROBLEM TO BE SOLVED: To provide a method which can automatically execute different tasks by manual work which is complicate and requires a lot of time. SOLUTION: The method includes a step for detecting the border between individual takes and a step for previously selecting a candidate region in the image of the video sequence succeeding or preceding to the detected border prior to the step for classifying the candidate regions. Further, it includes a step for merging the information of the continuous images to a single and coherent information aggregate and a step for detecting the change which informs a point peculiar to the phenomenon in the appearance of the token after the reliance analysis execution step. COPYRIGHT: (C)2005,JPO&NCIPI