| 摘要 |
<p>Gray pixel values, where the value of the pixel has one of a relatively large number of input levels, are directed through an error diffusion quantization process, for the conversion to one of a relatively small number of values. Each pixel of the gray level data (18) is modified (10) by a correspondingly weighted error correction term (20,22,24,28) or terms from previously processed pixels, generating a modified pixel value (12). This modified pixel value is compared (14) to a threshold value, the threshold value determined adaptively (15), based on 1) a dither pattern, having a dynamic range greater than 1, and 2) the value of the pixel to be thresholded, and optionally on its predetermined neighborhood, multiplied by a value representing an edge enhancement factor. The difference value between the modified pixel value (12) and the output pixel value (16) is distributed in accordance with a weighted distribution scheme to a set of neighboring unprocessed pixels, increasing or decreasing the gray values of such pixels. <IMAGE></p> |
