| 摘要 |
<p>PURPOSE: To fast and also precisely compress a data sample with a condition that a memory space is less than a conventional method. CONSTITUTION: First, 2nd and 3rd sample dots A, B and C are selected from plural pixels which constitute an actual image, line segments AB and BC are respectively defined from A and B, and B and C, and a maximum luminance error threshold is defined. A luminance error is calculated by using an actual luminance value and a compensated luminance value according to an algebraic logic, and it is decided whether the luminance error exceeds the threshold. When it does, the sample dots are moved, and after the sample dot A is set so that the sum of errors of the segments AB and BC is smaller than a total error, it is moved to the next sample dot, and an unequal sampling and compensation are performed. In this NSI system, e.g. in the case of 8:1 compression ratio, even though a compounding speed becomes 48 times as fast as a DCT compression, it can acquire a high quality image which has little deterioration in image quality.</p> |
