摘要 |
<p>An image is first divided into image objects. The dividing is done by using segmentation surfaces when the image is not compressed, while a demultiplexer is used for the dividing when it is compressed. Shape features are extracted from respective divided objects in terms of time elapsed. The extracted shape features are combined and time resolutions are time-wise determined for respective objects. Then, time resolutions are used for encoding or trans-coding various image objects as compressed output images.</p> |