Error diffusion is an important procedure for image and hologram quantization. The spatial distribution of the spectrum of the quantization noise is shaped by a filter function, which depends on the diffusion weights. The customary weights applied during the whole quantization process are optimized to yield the desired filter functions. Stability restrictions limit the optimization process. We suggest optimizing new weights locally after the quantization of each pixel. In this way the eventual occurrence of instabilities can be counteracted and the quality of the filter function improved.
© 1998 Optical Society of AmericaFull Article | PDF Article