摘要 |
PROBLEM TO BE SOLVED: To provide an image processing device capable of reproducing faithfully a three-dimensional model of an object by using a dynamic image photographed in an actual environment. SOLUTION: A background differentiation part 14 does not include a foreground, and generates a differential image by differentiating a background from a processing object image. A domain classification part 15 to a foreground domain extraction part 20 classify the processing object image into a first domain which is a background having high reliability, a second domain which is a background having low reliability, a third domain which is a foreground having low reliability, and a fourth domain which is a foreground having high reliability based on the differential image. When a projection point acquired by projecting one voxel to a processing object image photographed from one camera viewpoint belongs to the first domain, and a projection point acquired by projecting the voxel to a processing object image photographed from the other camera viewpoint belongs to the fourth domain, a three-dimensional model generation part 21 excludes from determination objects, the projection point projected to the processing object image photographed from one camera viewpoint, and generates the three-dimensional model of the object by a visual volume intersection method by using the residual projection point. COPYRIGHT: (C)2009,JPO&INPIT
|