| 摘要 |
PROBLEM TO BE SOLVED: To efficiently acquire a stable correspondence result which hardly falls into local solution at high speed when associating images. SOLUTION: A similarity image compression processing part 14 applies maximum value compression stepwise and repeatedly to a similarity image. An accumulation processing means 17, a similarity image update part 18 and a corresponding point determination part 19 perform corresponding point search processing to the similarity image to which the maximum value compression has been applied, and specify areas in which the corresponding points exist in each similarity image. A shift correction part 16 applies shift correction so that the areas specified that the corresponding points exist in each similarity image come to the center of the image. After the accumulation processing part 17, the similarity image update part 18 and the corresponding point search processing part 19 have performed the corresponding point search processing exclusively to the areas in each similarity image, the corresponding points in each similarity image are determined according to the amount of accumulation of the shift. COPYRIGHT: (C)2005,JPO&NCIPI
|
