Relation between grayvalue and noise frequencies in two- and three-level digital halftoning

In digital halftoning a close relationship exists between the local grayvalues in the input image and the spatial frequencies of the quantization noise of the quantized image. Halftoning algorithms usually reproduce low frequencies well by introducing quantization noise of higher frequencies. We present and discuss a simple model for the relationship between grayvalue and quantization noise frequencies. Optimization experiments with two and three quantization levels are presented.