| 摘要 |
<p><P>PROBLEM TO BE SOLVED: To achieve highly accurate camera-shake correction and video encoding. <P>SOLUTION: A CCD 18 provided with a color filter 36 takes three images, and a CMOS 21 with no color filter takes thirteen images. Three of the thirteen images are taken at the same time as imaging of the CCD 18. Motion vector data between respective images is calculated from the data of the thirteen images of the CMOS 21, the motion vector data is further calculated, motion vector data between a previous photographic image and a present photographic image, and motion vector data between the present photographic image and a following photographic image are calculated. A predicted image of the present photographic image is generated from the motion vector data, the previous photographic image and the following photographic image. A weighted average is calculated between the predicted image and the present photographic image, to generate a composite image with less blur. <P>COPYRIGHT: (C)2009,JPO&INPIT</p> |
