Simulated structured light 3D scanner datasets

This dataset contains a collection of simulated structured light 3D scanning scenes generated using the pbrt-v4 renderer. The controlled virtual environment enables consistent and repeatable experiments. The dataset includes multiple scenarios captured at varying camera and projector resolutions and with different camera models, including perspective, realistic, and realistic models with dispersion. This resource is designed to support research in 3D reconstruction, data processing, and machine learning, as well as benchmarking and validation of algorithms.

Each directory, corresponding to a specific camera–projector resolution and camera model (perspective, realistic, and realistic with dispersion), contains the following types of datasets:

• geometry calibration dataset consisting of six images of a fully illuminated calibration board captured in different positions

• phase calibration dataset including four full scanning passes of a calibration plate, composed of phase-shifted sinusoidal and Gray code patterns, with the number of patterns depending on the projector resolution

• object scan dataset containing scans of a target object different from the calibration plate, acquired through six passes of sinusoidal pattern and Gray code patterns, with the number of is determined by the projector resolution

Parameters used for projected images are as follows:

• Sine : period - 6 pixels, shift count - 6
• Gray : first image stripe width - 6 pixels and doubling for each next image

Marker positions are arranged on a grid defined relative to the calibration board center at (0, 0). Along the X-axis, markers span from –165.0 mm to +165.0 mm with a step of 55.0 mm for a total of 7 points. Along the Y-axis, markers span from –110.0 mm to +110.0 mm with a step of 55.0 mm for a total of 5 points. All markers lie on the plane resulting in 35 calibration points. 

Additionally, detailed instructions for dataset generation are provided in the README.md file, allowing users to reproduce the data acquisition setup or generate new datasets under customized conditions.