Implementation of Real-Time Stereo Matching System Based on Speckle Structured Light
Authors:
- 22 May 2021
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1537-1542
Abstract
Real-time and accurate acquisition of target depth information is the prerequisite for subsequent 3D reconstruction and abnormal behavior detection. For this application, this paper designs and implements a binocular stereo real-time system based on speckle. The main hardware components are binocular and speckle projector, and the main algorithm is (Semi-Global Matching) SGM's GPU implementation. The detection range of the system is 1.2m-6m, the frame rate can reach 30, the parallax level is 256, and the image resolution can reach 1280×720 pixel.Keywords
Funding Information
- National Natural Science Foundation of China (U1613214)
- Fundamental Research Funds for the Central Universities of China (N170402008)
This publication has 8 references indexed in Scilit:
- Embedded Real-time Stereo Estimation via Semi-global Matching on the GPUProcedia Computer Science, 2016
- A stereo matching algorithm based on census transformation and dynamic programming2014
- Real-time semi-global matching disparity estimation on the GPU2011
- Real-time Semi-Global Matching on the CPU2010
- Stereo Processing by Semiglobal Matching and Mutual InformationIEEE Transactions on Pattern Analysis and Machine Intelligence, 2007
- Evaluation of Cost Functions for Stereo Matching2007
- Accurate and Efficient Stereo Processing by Semi-Global Matching and Mutual Information2005
- A Taxonomy and Evaluation of Dense Two-Frame Stereo Correspondence AlgorithmsInternational Journal of Computer Vision, 2002