BVI-LOWLIGHT
收藏资源简介:
BVI-LOWLIGHT数据集由布里斯托大学视觉信息实验室创建,包含31800张14位图像,覆盖20个场景,由Nikon D7000和Sony A7SII相机在不同ISO设置下拍摄,旨在为低光条件下的图像去噪算法提供训练和基准测试。数据集通过一系列后处理步骤,包括暗帧移除、异常值检测和强度对齐,生成伪地面实况图像。该数据集特别适用于评估和改进在低光环境下工作的图像去噪技术,解决传统数据集在低光噪声特性表示上的不足。
The BVI-LOWLIGHT dataset was developed by the Vision and Information Laboratory of the University of Bristol. It comprises 31,800 14-bit images across 20 distinct scenes, captured using Nikon D7000 and Sony A7SII cameras under various ISO settings. This dataset is designed to support both the training and benchmarking of image denoising algorithms for low-light conditions. It generates pseudo-ground truth images via a series of post-processing steps including dark frame removal, outlier detection, and intensity alignment. Specifically, this dataset is tailored for evaluating and advancing image denoising techniques operating in low-light environments, addressing the limitations of traditional datasets in characterizing low-light noise properties.

- 1A Topological Loss Function: Image Denoising on a Low-Light Dataset布里斯托大学视觉信息实验室 · 2023年



