Real-Time Exposure Correction via Collaborative Transformations and Adaptive Sampling

Most of the previous exposure correction methods learn dense pixel-wise transformations to achieve promising results, but consume huge computational resources. Recently, Learnable 3D lookup tables (3D LUTs) have demon-strated impressive performance and efficiency for image enhancement. However, these methods can only perform global transformations and fail to finely manipulate local regions. Moreover, they uniformly downsample the input image, which loses the rich color information and limits the learning of color transformation capabilities. In this paper, we present a collaborative transformation framework (CoTF) for real-time exposure correction, which integrates global transformation with pixel-wise transformations in an efficient manner. Specifically, the global transformation adjusts the overall appearance using image-adaptive 3D LUTs to provide decent global contrast and sharp details, while the pixel transformation compensates for local context. Then, a relation-aware modulation module is designed to combine these two components effectively. In addition, we propose an adaptive sampling strategy to preserve more color information by predicting the sampling intervals, thus providing higher quality input data for the learning of 3D LUTs. Extensive experiments demonstrate that our method can process high-resolution images in real-time on GPUs while achieving comparable performance against current state-of-the-art methods. The code is avail-able at https://github.com/HUST-IAL/CoTF.

1. Introduction

Exposure correction [1] is a fundamental problem in the field of computational photography and computer vision, and has been extensively studied over the last few decades. Its purpose is to automatically correct over- or underexposed images taken under undesirable lighting conditions. Exposure correction plays an important role in many applications such as autonomous driving [14] and video understanding [3]–[8]. Figure 1.

Comparison of different transformation methods. (a) shows the computational effort of the different methods. We can see that the computational effort of the pixel transformation method SID increases significantly with resolution, while our method remains efficient at high resolution. (b) shows the pixel mapping relations for different transformations. 3D LUT performs a fixed global transformation based on pixel values, resulting in some unsatisfactory local contrast. While our method considers the pixel context and yields favorable results.