【摘要】 弱光图像增强旨在使隐藏在黑暗中的信息可见,以提高图像质量,在夜间目标检测和行为识别等计算机视觉任务中广泛应用。首先,从有监督和无监督两个角度出发,梳理了基于深度学习的弱光图像增强代表性算法,结合实现原理分析了其优缺点。其次,总结了常用的训练数据集和测试数据集。最后,讨论了目前已有算法存在的问题和未来可能的发展趋势。更多还原

【Abstract】 Low-light image enhancement aims to make information hidden in the dark visible to improve image quality and is widely used in computer vision tasks such as nighttime object detection and action recognition. Firstly, this paper compares representative algorithms for low-light image enhancement based on deep learning from both supervised and unsupervised perspectives. It analyzes their advantages and disadvantages concerning the implementation principles. Secondly,it summarizes the commonly used training datasets and test datasets. Finally, it discusses the problems of the existing algorithms and the possible future development trends.更多还原