Improved dot diffusion using optimized diffused weighting and class matrix

In this work, a high quality halftone image obtained by dot diffusion is proposed to reduce the deficiency gap with the error diffusion. Four kinds of filters with various sizes obtained by least-mean-square (LMS) are also introduced to simulate the human visual system (HVS). These filters are employed in the optimization procedures for class matrix of size 8x8. According to numerous of simulations, an optimized diffused weighting is determined. Many well-known halftone methods, which include direct binary search (DBS), error diffusion, ordered dithering, and previous dot diffusion are also involved for comparisons. As demonstrated in the experiments, the quality of the proposed dot diffusion is close to some error diffusion and is even superior to the well-known Jarvis and Stucki error diffusion or Mese's dot diffusion. Moreover, the dot diffusion inherently has the parallel processing advantage, which provides much higher executing efficiency than DBS or error diffusion.