Abstract
Existing image contrast enhancement methods are typically designed for specific tasks such as under-/over-exposure correction, low-light and backlit image enhancement, etc. The learned models, however, exhibit poor generalization performance across different tasks, even across different datasets of a specific task. It is important to explore whether we can learn a universal and generalized model for various contrast enhancement tasks. In this work, we observe that the common key factor of these tasks lies in the need of exposure and contrast adjustment, which can be well-addressed if high-dynamic range (HDR) inputs are available. We hence collect 46,928 HDR raw images from public sources, and render 328,496 sRGB images to build multi-exposure sequences (MES) and the corresponding pseudo sRGB ground-truths via multi-exposure fusion. Consequently, we train a network to generate an MES from a single sRGB image, followed by training another network to fuse the generated MES into an enhanced image. Our proposed method, namely UNiversal Image Contrast Enhancer (UNICE), is free of costly human labeling. However, it demonstrates significantly stronger generalization performance than existing image contrast enhancement methods across and within different tasks, even outperforming manually created ground-truths in multiple no-reference image quality metrics. The dataset, code and model are available at this https URL.
Abstract (translated)
现有的图像对比度增强方法通常针对特定任务（如欠曝/过曝校正、低光和逆光图像增强等）设计。然而，这些通过学习得到的模型在不同任务之间以及同一任务的不同数据集上的泛化性能较差。探索能否学习一个适用于各种对比度增强任务的通用且泛化的模型变得尤为重要。 在这项工作中，我们观察到这些任务中的共同关键因素在于需要调整曝光和对比度，如果能够使用高动态范围（HDR）输入，则可以很好地解决这一问题。因此，我们从公开来源收集了46,928张HDR原始图像，并生成了328,496张sRGB图像来构建多曝光序列（MES），并借助多曝光融合技术生成相应的伪sRGB地面实况数据。 接着，我们训练了一个网络以从单个sRGB图像中生成MES，随后又训练另一个网络将生成的MES融合成增强后的图像。我们的方法名为UNiversal Image Contrast Enhancer (UNICE)，无需昂贵的人工标签，并且在不同任务之间以及同一任务的不同场景下，其泛化性能显著优于现有图像对比度增强方法，在多个无参考图像质量指标上甚至超过了手动创建的真实地面实况数据。 该数据集、代码和模型可在提供的URL处获取。
URL
https://arxiv.org/abs/2507.17157
