LA-HDR: Light Adaptive HDR Reconstruction Framework for Single LDR Image Considering Varied Light Conditions

The high dynamic range (HDR) image recovery from the low dynamic range (LDR) image aims to estimate HDR image by decompressing luminance range and enhancing details of the LDR input. In practical usages, when faced with the over-exposed, the under-exposed or the low-light images, the state-of-art prediction methods lack the capability for ideally handling them. Aiming for this, a light adaptation HDR recovery framework (LA-HDR) is proposed, which includes the multi-images generation for adaptive details amplification in different light ranges, and the following multi-details fusion. To create the multi-images, first, the designed bit-depth enhancement network (EnhanceNet) produces the high bit-depth result with enhanced contrast. This result can be furtherly processed by user-defined denoising method to refrain the low-light noise. Meanwhile, the proposed exposure bias network (EBNet) estimates the global exposure bias of the input for rectifying the mid-range details. With the enhanced result and the exposure bias, the designed transfer functions adaptively create three multi-images containing the enhanced details in different light ranges, and they are fused by the designed multi-images fusion network (FuseNet) for the final HDR prediction. The amplification and fusion scheme ensures robust HDR recovery under different light conditions, eliminating high-light recovery artifacts from previous methods. The proposed fusion masks generation (FMG) and the global feature embedding (GFE) modules in FuseNet help eliminate the fusion artifacts. Experimental results show that LA-HDR acquires the best average performance under various light conditions, and it receives low influence from the input light conditions among the tested state-of-art HDR recovery methods.

I. Introduction

Due to the wilder luminance range and vivider color display, the high dynamic range (HDR) technology has been acknowledged to a wide range of applications, like cellphone photo shooting [1], [2], video games [3], movies [4], [5] and medical imaging [6]. In order to obtain the HDR image from the prevalent low dynamic range (LDR) camera, generally, the industry has developed two types of HDR image creation approaches: the multiple images fusion and the single image based HDR prediction, which is also called the inverse tone mapping (iTM). In the first approach, a series of LDR images are captured, representing the differences of the exposure [7], the ISO [8] or the noise attendence [9] of the scene, and then merged into a HDR image. The merge procedure is conducted either by deep neural networks [7], [10] or by weighted summation [11], [12]. Unfortunately, all these methods tend to deliver unpleasing ghost artifacts when the captured images contain fast-moving objects. In the second approach, the HDR image is estimated from only one LDR image using various mathematical models, like the gamma expansion [13], [14], the guided polynomial range expansion [15], [16] or the deep neural networks [17], [18]. All these methods assume that the input LDR images are in the appropriate light condition.