| 摘要 |
<P>PROBLEM TO BE SOLVED: To provide a virtual viewpoint image synthesizer capable of suppressing a depth estimated error and synthesizing a high quality virtual viewpoint image even when mapping between images is difficult. <P>SOLUTION: A depth estimation part acquires a camera parameter and image data from a camera image/camera parameter storage part and outputs a depth estimation result. At the time, as for a pixel for which the depth estimation is difficult, a subject to which the pixel belongs is estimated and a correction is performed using the depth information on the subject. When mapping is difficult as for a pixel at a portion where a depth in the vicinity of a boundary of the subject is rapidly changed, the depth information on the same subject is given. An image synthesizing part outputs a composite image viewed from a virtual viewpoint based on the camera parameter, the image data and the depth estimation result. <P>COPYRIGHT: (C)2013,JPO&INPIT |
