| 摘要 |
<p>PURPOSE: To provide the method for automation for generating an image for three- dimensional display with high precision by extracting the depth information of an original image to be the object of processing from that image and generating the image for threedimensional display according to this information. CONSTITUTION: The motion information of an object contained in the image is extracted. On this stage, the information is still handled as two-dimensional information. A display picture is taken on a coordinate plane and the motion of the object on this picture is described by a two-dimensional coordinate. Next, the relation of, correspondence between video frames is detected for grasping the motion of the object. Plural representative points are previously set to a video frame at time (t) [frame (t)] as a reference frame, the correspondent points of the respective representative points in the object frame at other time t' are pursued, and two-dimensional motion information is extracted. The relation of relative positions at every time is discriminated from the actual motion information of the object, and the depth information of each section is found from that relation. Parallax is decided corresponding to the depth information and images for both right and left eyes are generated. Generally, the closer the object is positioned, the more the parallax is enlarged. The right and left images are generated so that the parallax different for each section can not be generated.</p> |
