A multi-modeling electro-optical transfer function for display and transmission of high dynamic range content

Abstract

This paper proposes a multi-modeling Electro-Optical Transfer Function (EOTF) enabling efficient use of the code word to preserve the contrast details in the quantized High Dynamic Range (HDR) content. Different from the classic EOTF that employs one single model, the proposed EOTF considers four models to adjust the reproducible contrast according to the luminance level of the HDR content. Furthermore, the proposed EOTF and its inverse process functions can be integrated on HDR cameras and displays with a low computational cost. In order to evaluate the performance of the proposed EOTF, it has been applied into a HDR video transmission chain based on the High Efficiency Video Coding (HEVC) codec. The standardized Perceptual Quantizer (PQ) EOTF and Hybrid Log-Gamma (HLG) Opto-Electronic Transfer Function (OETF) are also introduced and compared with the proposed EOTF in terms of the reproducible contrast. The objective evaluation results show that the proposed EOTF brings a better quality of the decoded HDR video than that generated by the standard function. A subjective test was also conducted in a commercial HDR television and the results show that the proposed EOTF can present more visible details than PQ, particularly for the HDR content with dark scenes. Therefore, the proposed multi-modeling EOTF is useful for the transmission and display of the HDR content in the consumer electronics.