| 摘要 |
<p>An inputted original image (100) is divided into a 2nd block image by a 2nd block generating unit (2) to extract a 2nd feature by a 2nd feature extraction unit (6), and the 2nd block image and the 2nd feature are stored in an image memory unit (3). A control unit (15) reads only the block images having the same feature as that of a 1st block image from a 1st feature extraction unit (5) out of the image memory unit (3), and the read-out images are converted by an image converting/generating unit (7). The converted block images (107) are transmitted to a similar block searching unit (4) and matched with the 1st block image (101) by the similar block searching unit (4). The block image which minimizes the difference between both the block images is searched, selected, and transmitted to an encoding/multiplexing unit (9). In iterative conversion enconding/decoding, a high encoding efficiency is ensured and a higher quality image can be produced while the similar block searching is carried out at a high speed.</p> |
