论文标题
夜间摄影渲染的浅摄像机管道
Shallow camera pipeline for night photography rendering
论文作者
论文摘要
我们介绍了一条摄像机管道，用于在弱光条件下呈现视觉上令人愉悦的照片，这是NTIRE2022夜间摄影渲染挑战的一部分。鉴于任务的性质，该目标是由专家摄影师在口头上定义的，而不是依靠明确的地面真相图像，我们设计了一种手工制作的解决方案，其特征是浅层结构和低参数计数。我们的管道利用局部光增强子作为高动态范围校正的一种形式，然后全局调整图像直方图，以防止被洗涤的结果。我们将图像deno的形象比例地应用于较黑暗的区域，在那里它更容易被感知，而不会失去更明亮的区域的细节。该解决方案在竞争中排名第五，基于深度卷积神经网络的偏好投票数量与其他条目相当。代码可在www.github.com/availablataccepter上获得。
We introduce a camera pipeline for rendering visually pleasing photographs in low light conditions, as part of the NTIRE2022 Night Photography Rendering challenge. Given the nature of the task, where the objective is verbally defined by an expert photographer instead of relying on explicit ground truth images, we design an handcrafted solution, characterized by a shallow structure and by a low parameter count. Our pipeline exploits a local light enhancer as a form of high dynamic range correction, followed by a global adjustment of the image histogram to prevent washed-out results. We proportionally apply image denoising to darker regions, where it is more easily perceived, without losing details on brighter regions. The solution reached the fifth place in the competition, with a preference vote count comparable to those of other entries, based on deep convolutional neural networks. Code is available at www.github.com/AvailableAfterAcceptance.
