| 摘要 |
PROBLEM TO BE SOLVED: To execute the three-dimensional(3D) recognition of an object and to calculate the relative 3D position of the object only from picture data without requiring the optical characteristics of a light source and an image pickup device and the 3D position information of space to be measured. SOLUTION: A perspective projection image pickup part 1 picks up the perspective projection image of an object having plural feature points P0 , P1 ,.... A positive projection image pickup part 2 picks up the positive projection image of the object having plural feature points. A two-dimensional(2D) coordinate acquiring part 3 acquires the 2D coordinates of feature points in the perspective projection image and the 2D coordinates of feature points in the positive projection image. A camera parameter calculating part 4 calculates camera parameters by using the 2D coordinates of the feature points acquired by the acquiring part 3. A 3D coordinate calculating part 5 calculates a relative 3D position among plural feature points by using the 2D coordinates acquired by the acquiring part 3 and the camera parameters. |
