| 摘要 |
<P>PROBLEM TO BE SOLVED: To enhance both discrimination and invariance in identification of whether or not a subject in an image is a predetermined identification object. Ž<P>SOLUTION: A feature-point feature-quantity extracting part 22 extracts feature-point feature-quantities of feature points in a plurality of generation images including a positive image and a negative image, divides a feature point area as an area having the center of a feature point in an angular direction and a distance direction, on the basis of the feature point, into a plurality of small areas, in each of a plurality of response images obtained by filtering the generation images by a plurality of filters different in characteristics, determines a statistical quantity of a pixel value in the small area, for each of the plurality of small areas, and sets the statistical quantities in the plurality of small areas obtained from the plurality of response images, for the feature quantities, as the feature-point feature quantities of the feature points. The present invention is applicable to the identification as to whether the subject in the image is the predetermined identification object or not. Ž<P>COPYRIGHT: (C)2010,JPO&INPIT Ž
|
