DenseMatch: a dataset for real-time 3D reconstruction

We provide a database aimed at real-time quantitative analysis of 3D reconstruction and alignment methods, containing 3140 point clouds from 10 scenes. These scenes are acquired with a high-resolution 3D scanner. It contains depth maps that produce point clouds with more than 500k points on average. This dataset is useful to develop new models and alignment strategies to automatically reconstruct 3D scenes from data acquired with optical scanners or benchmarking purposes with respect to state-of-the-art methods. The dataset contains two type of files: - ".npz" files are python readable files containing the point clouds and additional metadata to use. - ".tar.gz" is a single large zip file which contains the different folders labeled similarly to the pointclouds. Each folder contains the RGB-D images and the camera calibration parameters that were used to reconstruct the 3D data. These are also the original output produced by the 3D scanner used to generate the data. A Python reader for those files is also provided in the project repository which is linked in the notes.

本数据库面向三维重建与配准方法的实时量化分析，共包含来自10个场景的3140组点云（point cloud）。所有场景均通过高分辨率三维扫描仪采集获取。该数据库包含深度图（depth map），由其生成的点云平均单组点数超过50万。
本数据集可用于开发新型模型与配准策略，以基于光学扫描仪采集的数据自动完成三维场景重建，亦可用于对标当前最先进方法的基准测试。
本数据集包含两类文件：
- ".npz"文件：为Python可读格式文件，存储点云与配套使用的元数据（metadata）。
- ".tar.gz"文件：为单个大型压缩包，内含多个与点云命名格式一致的文件夹；每个文件夹均存储了用于重建三维数据的RGB-D图像（RGB-D image）与相机标定参数（camera calibration parameter），同时也是生成该数据集所用三维扫描仪的原始输出结果。
本附注中附带链接的项目仓库内，同时提供了用于读取上述文件的Python读取工具。