High Dynamic Range Image Reproduction Using a Visual Contrast Mapping Model

摘要

A model for high dynamic range (HDR) image reproduction is proposed in this research. The main concept behind this model is to perform image reproduction based on the equivalence of perceived visual contrast for every pixel. A relative perceived visual contrast (RPVC) function is derived in this research from paper published by Burkhardt et al. to provide the base for visual matching between the HDR scene and the reproduced image in low dynamic range (LDR) medium. The adapting background luminance and physical contrast are the control parameters for this function. Many crucial features (for example bilateral type filter and !PT color spacing) found in prior HDR models (like iCAM06 etc.) are also incorporated in this model. Psychophysical experiment was performed in a specially configured setting to compare a real HDR scene with reproduction images on LDR monitor. The reproduced LDR images by six other HDR models were compared by pair comparison method. The results indicate that while iCAM06 has better performance in overall and bright region, this RPVC model has better performance in the dark region.