| 摘要 |
PROBLEM TO BE SOLVED: To provide an information processor, an information processing method, and a program which increase matching accuracy between an input image and a model image. SOLUTION: An input image 82 is matched with not only the model image 81 but N self-modulation images 91-1 to 91-N obtained by being converted by using each of N conversion coefficients from the model image 81. By this matching, one or more matching pairs between object feature points on the input image 82 having similar local feature quantity and model feature points on any of the model image 81 and the self-modulation images 91-1 to 91-N are obtained. Among the matching pairs, as for the matching pairs for the model feature points of the self-modulation images 91-1 to 91-N, in positional information of the model feature points, positions of corresponding points of the original model image 81 corresponding to the model feature points on the self-modulation images 91-1 to 91-N are coupled with the positions of the matched object feature points of the input image 82. The present invention is applicable to an image recognition device. COPYRIGHT: (C)2008,JPO&INPIT
|
