FACE EXPRESSION AMPLIFICATION DEVICE, EXPRESSION RECOGNITION DEVICE, FACE EXPRESSION AMPLIFICATION METHOD, EXPRESSION RECOGNITION METHOD AND PROGRAM

摘要

PROBLEM TO BE SOLVED: To correctly recognize even fine expressions, and to estimate the presentation strength of expressions.;SOLUTION: An input part 1 outputs the displacement from a non-expression time of the coordinate values of a plurality of feature points arranged on the face of an object person. An expression amplification parameter learning part 2 learns a parameter for amplifying expression data in each but the maximum presentation strength into expression data corresponding to the maximum presentation strength for each category. An expression amplification part 3 generates a plurality of hypotheses of the categories of expressions and the values of presentation strength, assumes that the category and the presentation strength of each hypothesis are correct, and amplifies input data into data whose category has been presented at the maximum. When the input data is amplified with various strengths, an expression recognition part 4 determines the recognition result of the category based on how far it is similar to the learnt expressions in the maximum presentation strength.;COPYRIGHT: (C)2012,JPO&INPIT