Evaluation of contrast sensitivity functions for the formulation of quality measures incorporated in halftoning algorithms

A comparative study of different CSFs (contrast sensitivity functions) as to their suitability for the formulation of quality measures in halftoning is presented. The evaluation of the different CSFs is implemented by weighting with each CSF the frequency-domain halftoning error of 44 halftone images, and then comparing the resulting weighted errors with the results of a psychovisual test. Comparison with psychovisual results indicates that the following factors significantly improve the performance of the CSF: incorporation of information about the printing device and the viewing conditions in the quantitative metric; incorporation of information about the angular sensitivity of the visual system: and utilization of a lowpass CSF instead of a bandpass one.<>