| 摘要 |
<P>PROBLEM TO BE SOLVED: To simultaneously achieve the enlargement of corresponding extent of expressions or directions of the face which can be collated and the improvement of the collating accuracy in the face collation based on an face image. <P>SOLUTION: Regarding a specific facial image expressing the face of a specific person, the feature quantities of a first feature quantities group determined by analyzing a first face image group for learning, for example, only an expressionless front face allowing for the solid discrimination of the face and the feature quantities of a second feature quantities group determined by analyzing a second face image group for learning, for example, a facial image group configured of the faces whose expressions and directions are variable and allowing for the solid discrimination of the face are calculated and stored. For the input face image to be collated, the feature quantities of each group is calculated. The feature quantities for each group are compared, between the facial image to be collated and the specific face image, and their similarities are calculated, and are utilized in comprehensive manner and the collation judgment of both the facial images are carried out. <P>COPYRIGHT: (C)2008,JPO&INPIT |
