| 摘要 |
PROBLEM TO BE SOLVED: To select more feature points with high evaluation values even when detecting feature points by block division.SOLUTION: A feature point detection part 10 calculates evaluation values of respective pixels in each unit block obtained by dividing a digital image into blocks with a fixed size and detects candidate points to be candidates of feature points. A candidate point storage part 20 stores the candidate points detected by the candidate point detection part 10 and evaluation values of the candidate points. A feature point number determination part 30 determines the number of feature points in the unit block in each unit block on the basis of the evaluation values stored in the candidate point storage part 20. A feature point selection part 40 selects feature points corresponding to the number of feature points determined by the feature point number determination part 30 from the candidate points stored in the candidate point storage part 20 in each unit block. |
