Simulation of Image Detail Visibility using Contrast Sensitivity Functions and Wavelets

摘要

Natural images contain a lot of information imperceptible to the human eye. While it is of key importance to maintain the visible information in image reproduction processes, the imperceptible information can, with good benefit, be discarded. In order to identify the imperceptible information, a method for simulating this particluar aspect of the human visual system is required. In this paper we present a robust contrast filtering method for simulating the detail visibility of natural images. The method removes image spatial-frequency components that are undetectable at a given viewing distance. The method is based on the relationship between contrast sensitivity and spatial frequency (contrast sensitivity functions) and octave-wise spread over the spatial frequency range matched by wavelet decompositions. An experimental evaluation of the method, where simulated distance is compared to observered discrimination distance, shows promising results. The proposed contrast filtering method has a wide range of applications; it can be used in, e.g., image quality metrics, image compression, gamut mapping, and halftoning.