| 摘要 |
PROBLEM TO BE SOLVED: To realize high compressibility, at the same time to enable the automatic classification of characteristic points and to promote an automatic collation system by deciding the phase map structure of an image, eliminating singular points in the phase map structure and separately encoding the significant point and a continuous phase map structure. SOLUTION: A 1st input image 50 forms a phase map 80, including a residual phase 82 and a singular point 84. The phase 82 is compressed through appropriate processing, and a compressed residual phase 86 is obtained. The point 84 and the phase 86 forms a compressed record 88 of the image 50 and can be stored in a database. When collation between the image 50 and a 2nd input image 52 is desired, the record 88 and a compressed record 90 corresponding to the image 52 carry out the collation of the combination of respective singular point information and the residual phase at a singular point position, without conducting compression concellation of either or both. Thus, collation processing can be accelerated drastically. |
