RI: Small: Low Cost Technologies to Improve the Quality of 3D Scanning

Brown University, Providence RI

Investigators

Abstract

This project develops a new mathematical framework, algorithms, and optical designs to increase the resolution and precision of line and area 3D scanners, and to minimize the hardware cost. Combining simple optical design techniques and novel mathematical formulations, this project builds the next generation of low cost high resolution 3D scanner systems based on active illumination and 3D triangulation. The project designs, fabricates, and demonstrates prototype 3D scanners based on the new technologies. The project integrates research with education. Courses taught by the principal investigator on a regular basis provide the necessary mathematics, software, and practical details to design and build 3D scanners using off-the-shelf low cost semiconductor lasers, small LED projectors, and web-cams, as well as digital geometry processing software. Undergraduate and graduate students learn in these courses to design and build low cost 3D scanners based on the technologies being developed in this project. By significantly lowering the cost of high resolution 3D scanning systems, the technology, software and methods being developed in this project have significant impact in the multiple applications in computer vision, computer graphics, and robotics. The research studies a new mathematical formulation for 3D optical triangulation based on interval arithmetic, where 3D points are only determined within certain bounds along the camera rays, and multiple measurements are used to tighten these bounds. The project introduces the "Line Segment Cloud" as an alternative surface representation, and as a tool to visualize the measurement errors within the proposed framework. The project develops new variational surface reconstruction methods with line segment constraints tailored to line segment clouds. Enabled by the new formulation, the project is constructing: 1) novel laser line and structured light super-resolution 3D scanners, where the projectors are mounted on small displacement motion platforms; 2) structured light area 3D scanners where the different 3D patterns are projected by fixed non-coaxial LED slide projectors, and neither DLP or LCD projector engines are used; and 3) super-resolution 3D scanners based on volume carving, optimized for existing line and area based 3D scanners, where camera and projector are rigidly mounted with respect to each other. The project is also developing techniques to design laser line projectors optimized for optical triangulation using inexpensive optical components, where the thickness of the projected line is significantly reduced and the depth of field is increased. The project develops new calibration methods and performance metrics to evaluate these technologies.