A Pattern and Calibration Method for Single-Pattern Structured Light System
A Pattern and Calibration Method for Single-Pattern Structured Light System
复制标题
一种单图案结构光系统的图案及标定方法
DOI：
10.1109/tim.2019.2929281
复制
发表时间：
2020-06-01
影响因子：
5.6
通讯作者：
Song, Zhan
中科院分区：
文献类型：
--
作者：
Feng, Zexi;Man, Duhu;Song, Zhan
结构光是主动立体视觉中广泛采用的一种轮廓测量方法。最近的方法已经开发了基于学习的单模式结构光系统和非结构光立体视觉系统。然而，他们的方法需要额外的相机或衍射光学元件。此外，基于学习的结构光系统的标定方法需要大量的图像，而立体视觉方法不能检测到被遮挡的区域。传统上，数字投影仪被视为伪相机，并且条纹图案被用来测量深度。然而，在基于条纹的方法中，采用附加图案来校准相机-投影仪系统。本文提出了24个可靠特征和一种新的标定算法。该算法不需要额外的条纹图案就可以对条纹图案进行标定，并且利用RGB特征，大大增加了编码列数。实验结果表明，该方法具有较好的准确性和鲁棒性.
Structured light is a widely employed approach for profile measurement in active stereo vision. Recent methods have developed learning-based single pattern structured light systems and unstructured light stereo vision systems. However, their methods require additional cameras or a diffractive optical element. In addition, the calibration methods of the learning-based structured light systems require a large number of images, while the stereo vision method cannot detect the blocked region. Classically, a digital projector is treated as a pseudocamera, and the stripe pattern is employed to measure the depth. However, an additional pattern is employed to calibrate the camera-projector system in the stripe-based method. In this paper, 24 reliable features and a novel calibration algorithm are proposed. With the proposed calibration algorithm, the stripe patterns are calibrated without additional patterns, and with the proposed RGB features, the number of encoded columns is enlarged significantly. Experiments demonstrate that the proposed method is accurate and robust.