| 摘要 |
High addressability image signals, each representing a pixel in an image, where the signal has one of a relatively large number of signal levels, are directed through an error diffusion quantization process, for the conversion of the signal to one of a relatively small number of levels. Each signal is modified by a correspondingly weighted error correction term signal or signals from previously processed image signals, thereby generating a modified image signal. This modified image signal is compared to a threshold to derive a preliminary output signal having a smaller number of levels. The preliminary output signal is directed to an output if it has the same level as the previous output signal. If the preliminary output signal does not have the same level as the previous output signal, the preliminary output signal is directed to the output only if the previous output signal was one of a contiguous group of a preselected number of like level signals. Otherwise, the previous output value is directed to the output. The actual output signal is then compared to the modified input signal to derive an error signal for the diffusion feedback loop that maintains gray density.
|
