| 摘要 |
PROBLEM TO BE SOLVED: To properly detect an area including a moving body and/or an area including no moving body, in a frame of a video image captured from a moving point-of-view.SOLUTION: An image processing apparatus includes: a motion deriving unit which derives a target vector field representing motion of various local areas in a frame of a target video image, from the video image; an approximation unit which approximates the target vector field as a linear combination of a plurality of reference vectors; and an area detection unit which detects an area including a moving body and/or an area including no moving body in the frame of the target video image, according to the size of a component of a difference vector field, which is a difference between the target vector field and the linear combination. The reference vector fields are eigenvector fields obtained as a result of performing principal component analysis on test vector fields after deriving the test vector fields representing the motion of various local areas in the frame of a test video image, from the test video image formed by capturing a test space while moving a point of view in various directions. |
