| 摘要 |
<p>A 16 x 16 image block of an HDTV image is converted by means of a discrete cosine transformation according to a decision process into a 16 x 16 full-image block or separately into two 16 x 8 half-image blocks. Image value differences in half-image point pairs of two image lines of the same half-image are added together absolutely to form a half-image sum. Likewise, image value differences in full-image point pairs of two image lines of the full image are added together to form a full-image sum. The full-image sum multiplied by a full-image evaluation factor is subtracted from the half-image sum to give a positive decision result in some cases. If the decision result is positive, the full image is converted; otherwise, both half-images are separately converted.</p> |
