Efficient-HDRTV: Efficient SDR to HDR Conversion for HDR TV

The existing deep neural network (DNN) based SDR (Standard dynamic range) to HDR (High dynamic range) conversion methods outperform conventional methods, but they are either too large to implement on a device or with quantization artifacts generated on smooth regions on an image. We propose an efficient neural network for the SDR to HDR conversion, namely "Efficient-HDRTV". It consists of two efficient structures GIM (Global Inverse Mapping) and LIM (Local Inverse Mapping). The key features of GIM and LIM use the small series of a basis function with its coefficient function, which are implemented using small number of convolutions, logarithm and exponential functions. They are combined with other convolutional layers so that the entire network can be jointly trained for learning inverse tone, enhanced details and expanded color gamut from SDR to HDR. Thanks to the GIM and LIM, we can keep our network small with good performance. Our experimental results show that Efficient-HDRTV is much lighter but performs better than the state of the arts.

1. INTRODUCTION

High dynamic range (HDR) TV has been improved to be able to display upward of 2000 nits of peak brightness for HDR contents with much wider color gamut such as DCI-P3 [1] and BT.2020 [2]. With the improvement of TV, a demand for HDR contents has also increased. However, original HDR video contents are not enough to satisfy the demands, and SDR videos still dominate the market. For solving this issue, SDR to HDR conversion methods have been proposed to an industry. Conventional methods predict the inverse tone mapping curve between SDR and HDR images using image statistics. Recent deep neural network (DNN) methods train convolutional neural networks (CNNs) using a set of the paired SDR and HDR images to learn the relationship between them. The DNN methods [3], [4], [5], [6], [7] have demonstrated superior outputs than conventional methods. However, the existing methods use the large sized DNN, which is a very critical problem to implement the network in the display device such as a TV and a AR/VR device. In this paper, we propose a novel neural network, called "Efficient-HDRTV", which is a very light network comparing to the state of the arts for SDR to HDR conversion.