感知友好的低光照视频增强技术研究

Affected by the light conditions, low light videos are usually created in daily life. Due to the low visibility of global or local areas, much of the visual information are buried, which greatly increases the difficulty for users to analyze and interpret the video. Many low light video enhancement techniques have been proposed for improving the visibility. However, along with the development trend of popularization, socialization and entertainment in the video industry, the core demands of users for low-light video enhancement technology have shifted from the originally visibility enhancement to achieving physically realistic and visually pleasing results, which are not enabled by existing methods. To adapt this change in demand, we study perceptually-friendly low light video enhancement in this project, which includes the following three problems in detail: (1) perceptually-friendly visibility enhancement, (2) pure black area enhancement based on the spatio-temporal similarity, (3) adaptive texture-preserving spatial-temporal video denoising. The study of this project will provide new research perspectives and solutions for low light video enhancement, and expand the application of this technology in practice.

受光照条件影响，日常生活中常会产生低光照视频。此类视频全局或局部区域能见度通常较低，导致其中许多视觉信息被掩埋，极大地限制了用户对视频的后续分析与解读。为改善能见度，现有研究提出了诸多低光照视频增强方法。然而，伴随着视频产业大众化、社交化及娱乐化的发展趋势，用户对于低光照视频增强技术的核心诉求已从最初的能见度提升，转为现有方法无法胜任的，期望获得兼顾物理真实性及视觉愉悦性结果的更高层次需求。为适应这一需求变化，本项目研究感知友好的低光照视频增强，具体包括三个研究内容：（1）感知友好的能见度增强，（2）基于时空相似性的纯黑区域增强，（3）自适应保纹理时空一致视频降噪。本项目研究将为低光照视频增强领域提供新的研究视角和解决方案，并扩大该技术在实际中的应用范围。

受光照条件影响，日常生活中常会产生低光照视频。此类视频全局或局部区域能见度通常较低，导致其中许多视觉信息被掩埋，极大地限制了用户对视频的后续分析与解读。为改善能见度，现有研究提出了诸多低光照视频增强方法。然而，伴随着视频产业大众化、社交化及娱乐化的发展趋势，用户对于低光照视频增强技术的核心诉求已从最初的能见度提升，转为现有方法无法胜任的，期望获得兼顾物理真实性及视觉愉悦性结果的更高层次需求。为适应这一需求变化，本项目研究感知友好的低光照视频增强，具体包括三个研究内容：（1）感知友好的能见度增强，（2）基于时空相似性的纯黑区域增强，（3）自适应保纹理时空一致视频降噪。本项目研究将为低光照视频增强领域提供新的研究视角和解决方案，并扩大该技术在实际中的应用范围。