The technology of 3D acquisition behind face recognition is structured light scanning (SLS). As one of most essential non-contact optical 3D measure methods, SLS has been broadly used in a variety of applications such as virtual reality, animation, and industrial inspection.
A typical SLS system consists of one camera and one projector. A decoding method is adopted to extract the pattern feature from the captured images and establish correspondences between the camera and the projector.
Researchers from the Shenzhen Institutes of Advanced Technology (SIAT) of the Chinese Academy of Sciences introduced an accurate infrared SLS system for dynamic 3D acquisition. The study was published in Applied Optics on May 15.
This infrared SLS system combined the Gray code with stripe shifting as the code strategy to obtain more precise results. It showed advantages of capturing pattern sequences at a speed of 29Hz, which could meet the framerate requirement of most real-time 3D applications.
Moreover, to assure that the SLS system could work at the maximized capturing speed, the team parallelized the decoding algorithm on the GPU device. Results showed that the proposed algorithm reduced the decoding time from 1.5s to 0.5ms.
To further improve the accuracy and the feasibility of the proposed SLS, the team put forward a practical framework to calibrate the system.
In the future, they plan to develop a full-view live 3D imaging system consisting of multiple infrared SLS devices.
Fig. (Left) A china vase; (Middle) 3D point cloud of the vase reconstructed with the LED based SLS; (Right) 3D point cloud of the vase reconstructed with the infrared based SLS. (Image by SIAT)
52 Sanlihe Rd., Beijing,
Copyright © 2002 - Chinese Academy of Sciences