CRF ESTIMATION BASED HDR IMAGE GENERATION METHOD

摘要

In this paper, we propose a single image based high dynamic range (HDR) imaging method. In the method, the inverse camera response function (CRF) is modeled according to the empirical model of CRF firstly; then the optimal inverse CRF is resolved and multiplied by a weighting function to make it smooth in the areas near the maximum and minimum pixel values; finally, the HDR image is generated by conducting the inverse CRF on the input single LDR image. Unlike most of the existing methods, the experimental results show that the proposed approach expands the image from both high and low luminance areas. The results of image quality metric indicate that the proposed method induces less contrast distortion and produces high visual quality HDR image compared with other recent methods.