Resolving the biases in threshold decomposition-based multitoning

Threshold decomposition is a technique commonly used to produce multitones in multilevel halftoning. It separates an input image nonlinearly into energy planes, halftones them sequentially with a binary halftoning algorithm, and finally combines the binary halftoning results to produce a multitone. As planes are handled sequentially from the brightest layer to the darkest layer under a stacking constraint, there are biases to favor the brighter layers and the darker features of the image. This in turn makes the resultant multitone impossible to report the original image ideally. This paper proposes a solution to eliminate these biases and improve the visual quality of a produced multitone.