摘要 |
PROBLEM TO BE SOLVED: To provide a free viewpoint image composition device capable of composing an image for a highly precise three-dimensional virtual scene directly from an actual video without requiring any procedure for finding a three-dimensional surface structure of an object. SOLUTION: This free viewpoint image composition device is provided with an actual video data capture part 1, an omunidirectional photograph network positioning part 2, an object area decision part 3, a visible face subarea division part 4, an actual image data calibration part 5, a visible face subarea association identification part 6, a morphing processing part 7, a background image composition part 8, and an image display part 10. By a series of processing consisting of image capturing, visible face division, character point association between the adjacent visible faces, related image morphing transformation, and composite scene rendering, an image of a three-dimensional virtual scene is composed directly from a two-dimensional structure data for the actual image from a plurality of viewpoints. COPYRIGHT: (C)2006,JPO&NCIPI |