Research on HDR Image Rendering Based on Image Appearance Effect Predicting

A new high-dynamic-range image rendering algorithm based on image appearance effect predicting was developed in this paper.The predicting models for "Hunt effect" and "Stevens effect" were integrated into HDR image rendering algorithm based on image color appearance model.Then,the best values for the modulating parameter "c" in the predicting model of "Hunt effect" and the modulating parameter "m" in the predicting model of "Stevens effect" were obtained respectively by the corresponding experiments aimed at the two kinds of purposes of evaluation(rendering preference evaluation and the evaluation for perceptual accuracy of reproducing the real-world scenes).The final predicting models for "Stevens effect" and "Stevens effect" were integrated into the flow of algorithm.The performance of the algorithm improved was evaluated by experiments of rendering preference and perceptual accuracy.The result of rendering preference experiment shows the algorithm improved has significant better performance than four algorithms in PhotoShopCS4 and the iCAM algorithm unimproved.The result of perceptual accuracy experiment shows the algorithm improved has significant better performance than the iCAM algorithm unimproved,and has the approximately equivalent performance to PhotoShopCS4 algorithms.Finally,the computing speed of algorithm was improved by discrete FFT and down-sampling technology.