LCDMoire
收藏资源简介:
LCDMoire数据集是由华为诺亚方舟实验室和瑞士苏黎世联邦理工学院联合创建,专为2019年国际计算机视觉大会(ICCV)上的图像去摩尔纹挑战而设计。该数据集包含10,200对合成图像,每对包含一张受摩尔纹影响的图像和一张清晰的参考图像。数据集的创建过程涉及复杂的图像生成管道,模拟了智能手机拍摄LCD屏幕时产生的摩尔纹效应。LCDMoire数据集主要用于评估和推动基于深度学习的图像去摩尔纹技术的发展,旨在解决数字摄影中摩尔纹对图像质量的影响问题。
LCDMoire dataset was jointly created by Huawei Noah's Ark Lab and ETH Zurich, and was specifically designed for the Image Demoiréing Challenge held at the 2019 International Conference on Computer Vision (ICCV). This dataset contains 10,200 pairs of synthetic images, where each pair consists of one moiré-contaminated image and one pristine reference image. The dataset construction involves a sophisticated image generation pipeline that simulates the moiré artifacts generated when smartphones capture LCD screens. The LCDMoire dataset is primarily used to evaluate and promote the development of deep learning-based image demoiréing techniques, with the objective of addressing the negative impact of moiré patterns on image quality in digital photography.




