| 摘要 |
A multi-level input image is divided into blocks, each of which is quantized to a bi-level output block. A single quantization error value is calculated for each block, and distributed over neighboring blocks that have not yet been processed. Blocks containing abrupt changes in image content are preferably converted by comparing each picture element individually with a threshold value. Blocks not containing abrupt changes are preferably quantized by comparing the average value of the block with multiple threshold values to determine a multi-level block output value, from which a predetermined pattern of bi-level picture elements is generated.
|
