Testing HDR Image Rendering Algorithms

Eight high-dynamic-range image rendering algorithms were tested using ten high-dynamic-range pictorial images. A large-scale paired comparison psychophysical experiment was developed containing two sections, comparing the overall rendering performances and grayscale tone mapping performance respectively. An interval scale of preference was created to evaluate the rendering results. The results showed the consistency of tone-mapping performance with the overall rendering results, and illustrated that Durand and Dorsey’s bilateral fast filtering technique and Reinhard’s photographic tone reproduction have the best rendering performance overall. The goal of this experiment was to establish a sound testing and evaluation methodology based on psychophysical experiment results for future research on accuracy of rendering algorithms.