| 摘要 |
<p><P>PROBLEM TO BE SOLVED: To detect a motion vector with a high encoding efficiency, high accuracy and less arithmetic amount even in the case of a motion compensation prediction wherein an interlacing scanning image is an object. <P>SOLUTION: When detecting a block similar to block data 101 of an encoded image with a frame format, the positions of block data of a reference image to be selected with the frame format are made to be at interval of four pixels in the horizontal direction from the data 101 in compliance with the frame format in a reference image region, and the positions are selected on a succeeding line by deviation from the positions selected on a preceding line by one pixel. Thus, the maximum distance between a vector searching point and a point whose searching is avoided in the frame format is equal to a√2 pixel and the maximum distance between a vector searching point and a point whose searching is avoided in a field format is equal to one pixel so that the maximum distance between the vector point not searched and the searched vector point is equal to√2 pixel. <P>COPYRIGHT: (C)2004,JPO&NCIPI</p> |
