A dynamic range adjustable inverse tone mapping operator based on human visual system

Conventional inverse tone mapping operator (iTMO) reconstructs high dynamic range (HDR) images with fixed dynamic range and tend to produce extended distortion in both high and low exposure regions of HDR images. This paper proposes a dynamic range adjustable inverse tone mapping algorithm based on single LDR image, which combined photoreceptor response and adaptation. Firstly, the linearized image is converted to retinal response in the LDR environment. Then, it is extended to obtain the retinal response corresponding to HDR scenes. This extension can also be adjusted according to the target dynamic range. Since the different states of light adaptation, expanded retinal response is converted into a set of HDR images with different exposure background intensity. The corresponding processing for different exposure regions can effectively reduce the distortion of high exposure and low exposure regions. Finally, this group of HDR images is synthesized base on the corresponding weighted graph. The efficiency and high visual quality of the proposed algorithm are validated in open-source datasets, and the superior performance of proposed algorithm is also proved by objective evaluations of different types of LDR images.