3D-printed objects dataset, fallen leaves dataset
收藏资源简介:
本文使用了一种新的方法,名为差分移动显示光度立体(DMDPS),该方法通过手机上的显示和摄像头来捕捉图像,并进行表面法线重建。研究中采用了两种数据集:一种是真实世界的3D打印对象,用于获取训练数据集的地面真相;另一种是落叶数据集,通过该方法揭示了重建的表面法线和反照率。3D打印对象数据集是通过捕获3D打印对象的基照明图像并使用相应的3D建模文件获得地面真实表面法线图来创建的。落叶数据集则是在野外对象上应用DMDPS,展示了从这些对象中重建的表面法线和反照率。
This paper presents a novel method termed Differential Mobile Display Photometric Stereo (DMDPS), which captures images via the display and built-in camera of a smartphone and conducts surface normal reconstruction. Two datasets are employed in this study: one is a real-world 3D printed object dataset used to acquire the ground truth for the training dataset, and the other is a deciduous leaf dataset, where the reconstructed surface normals and albedo are revealed through this method. The 3D printed object dataset is created by capturing base illumination images of 3D printed objects and obtaining ground-truth surface normal maps using their corresponding 3D modeling files. The deciduous leaf dataset, by contrast, applies DMDPS on field objects and demonstrates the reconstructed surface normals and albedo derived from these objects.




