Prediction of Print Defect Perception

This study examines the prediction of print defect perception (banding) of the human visual system (HVS) by combining detection probabilities of contrast components from wavelet and sinusoidal decompositions of defect patterns. Detection probabilities for basis function components are obtained from subjective tests. Prediction performance is presented in terms of the deviation of the contrast thresholds at the 92% detection probability obtained from subjective tests on the simulated print defect patterns. Results from tests run on 21 subjects indicate that prediction errors can be obtained on the order of the subjective test variability used to obtain detection probabilities. While the wavelet approach resulted in less systematic error than the sinusoidal approach, both performed about the same once the systematic error was removed. An advantage of the wavelet approach includes having a framework more conducive for modeling independent visual channels of the HVS and for obtaining efficient orthogonal decomposition of the defect patterns.