投影机
计算机视觉
校准
人工智能
重射误差
计算机科学
像素
像面
特征(语言学)
双眼视觉
相(物质)
摄像机切除
数学
图像(数学)
物理
语言学
统计
哲学
量子力学
作者
Ran Jia,Junpeng Xue,Wenbo Lu,Zeyu Song,Zhichao Xu,Shuxin Lu
出处
期刊:Sensors
[Multidisciplinary Digital Publishing Institute]
日期:2024-03-20
卷期号:24 (6): 1987-1987
被引量:1
摘要
Binocular structured light systems are widely used in 3D measurements. In the condition of complex and local highly reflective scenes, to obtain more 3D information, binocular systems are usually divided into two pairs of devices, each having a Single Camera and a Projector (SCP). In this case, the binocular system can be seen as Dual Cameras-Projector (DCP) system. In the DCP calibration, the Left-SCP and Right-SCP need to be calibrated separately, which leads to inconsistent parameters for the same projector, thus reducing the measurement accuracy. To solve this problem and improve manoeuvrability, a coupled calibration method using an orthogonal phase target is proposed. The 3D coordinates on a phase target are uniquely determined by the binocular camera in DCP, rather than being calculated separately in each SCP. This ensures the consistency of the projector parameters. The coordinates of the projector image plane are calculated through the unwrapped phase, while the parameters are calibrated by the plane calibration method. In order to extract sub-pixel accuracy feature points, a method based on polynomial fitting using an orthogonal phase target is exploited. The experimental results show that the reprojection error of our method is less than 0.033 pixels, which improves the calibration accuracy.
科研通智能强力驱动
Strongly Powered by AbleSci AI