提出一种基于柔性靶标定位实现图像拼接的多相机三维测量系统,采用一个激光投影仪投影大幅条纹,多相机分布式采集的方法扩展视觉三维测量系统的测量范围。标定过程首先使用小型平面靶标标定基准相机二维图像坐标和相位值到三维世界坐标的映射关系,之后在相邻相机部分视场(FOV)重合的前提下,利用柔性靶标定位标定相邻相机图像坐标的转换关系,最后将各个相机的图像坐标全部转化到基准相机的图像坐标系下完成图像拼接,由基准相机图像坐标到世界坐标的映射完成全局三维测量。实验结果表明,使用图像拼接方法的测量精度略低于相机单视场测量的精度,但精度损失较小,满足工业在线测量的要求。该方法避免使用昂贵的辅助测量仪器和加工高精度大型靶标,为多相机视觉测量提供了成本低、使用方便的解决方案。

A multi-camera three-dimensional (3D) measurement system using an image stitching method based on flexible calibration target positioning is proposed. In order to expand the measuring range of the 3D shape measurement system, a laser projector is used to project large fringe images, followed with distributed multiple cameras to grab images of each field of view (FOV). The first step of calibration process is to establish the mapping relationship from the image coordinates and absolute phase to the world coordinates by using the reference camera with a small planar calibration target. Then, with the FOV of adjacent cameras partially overlapping, the flexible calibration target positioning is applied to calibrate mapping relationships of the image coordinates from adjacent cameras. After that, the image coordinates of all the other cameras are converted to the image coordinates of the reference camera by the new image stitching method. Finally, the reference camera coordinates are transformed to the world coordinates. The experimental results show that the accuracy of this method is slightly lower than the local measuring method with a single camera. However, the accuracy loss is not so severe, meeting the requirement for industrial on-line measurement. This method does not require expensive auxiliary measuring instruments or manufacturing large calibration targets with high precision, thus offering a low-cost and easy alternative for multi-camera 3D shape measurement systems.