基于多靶标恒约束的柔性视觉测量系统现场高精度标定方法研究

As one of the difficulties in measurement research field, the flexible vision measurement system has been used widely in industrial environment. The project is aiming at the two key scientific problems in the flexible vision measurement system, one is "Decoupling and calibration method of parameters for single camera", and the other is "Field global calibration method of the flexible vision measurement system". This project is aiming at "high precision measurement", will develop the research in the whole process from the selection of cameras, planar target design, corner extraction, camera calibration to global field calibration of measurement system. This project will make the related basic theory and application technology research work by theoretical analysis, numerical simulation and experimental verification. Steady and high precision corner extraction method based on local area gray feature model will be researched in this project. Some constraints such as linear constraint and vanish point constrain will be introduced in the calibration of the calibration of single camera parameters, the radial and tangential distortion parameters will be separately calibrated firstly, and the decoupling of inner linear parameters and nonlinear distortion parameters will be achieved. High precision and field pose calibration of two cameras and multi-camera based on constant constraint will be researched, then global field calibration and optimization method for the flexible vision measurement system will be explored. The accuracy evaluation methods for corner extraction, the calibration of single camera parameters and global calibration for the flexible vision measurement system will be researched. Research results will enrich and develop the related measurement theory and method, and provide strong theoretical and technical support for the application based on the flexible vision measurement system.

柔性视觉测量系统在工业环境中应用广泛，是测量领域研究难点之一。本项目围绕柔性视觉测量系统中的"单相机参数的解耦标定方法"与"柔性视觉测量系统现场全局标定方法"两个关键科学问题，以"高精度测量"为目标，从相机选择、靶标设计、角点提取、相机标定、测量系统现场全局标定等方面入手，通过理论分析、数值仿真与实验验证等途径开展研究。研究基于局域灰度特征模型的角点稳定且高精度的提取方法；在单相机参数标定过程中引入直线约束与灭点约束等约束条件，先单独标定径向与切向畸变参数，实现相机内部线性参数与非线性畸变参数的解耦；从研究基于恒约束的两相机/多相机间位姿现场高精度标定方法出发，探索柔性视觉测量系统现场全局标定及优化方法；研究角点提取、单相机参数标定、柔性视觉测量系统全局标定的精度评价方法。项目研究将丰富和发展相关的测量理论和方法，为基于柔性视觉测量系统的应用提供理论和技术支撑。

提出了一种基于方形环模板和局域灰度对称因子的棋盘格角点亚像素定位算法，无需高斯滤波与梯度计算，可以很好地解决算法数据量大、运算速度慢等问题，有效降低噪声和其他不确定因素对角点亚像素定位精度的影响。.针对椭圆经透视投影后所得椭圆特征图像相对于椭圆中心具有几何对称的特性，提出了一种基于椭圆中心灰度对称的圆心定位算法。.由于张正友相机标定算法对靶标姿态的布局没有明确加以说明，在实际标定时可能会影响标定精度。尝试将均匀设计思想引入靶标姿态布局的设计中，取得了较好的效果。.针对柔性视觉测量系统的现场全局标定，设计了基于恒外参约束的双平面靶标的旋转标定架，用以标定相邻相机间相互位姿关系。研究并完善了利用双平面靶标进行柔性视觉测量系统全局位姿关系高精度标定的算法。.研制了相机外参高精度标定架和相关算法，利用其进行了柔性视觉测量系统全局标定的精度分析与评价。.基于上述研究成果，针对常见3D汽车四轮定位仪占用场地较大且更换相机不方便等缺点，研发了柔性自适应两相机3D四轮定位仪，该产品具有体积小、更换相机方便、维护简单、场地适应性强等优点。