为在惯性约束聚变系统中开展靶丸空间坐标的高精度测量,针对人工方式瞄准方法在安装调试上存在费力费时、自动化程度不高等缺点, 采用基于双目视觉的方法, 利用高精度2维标定板, 通过迭代方法使角点提取精度达到0.1个亚像素, 采用奇异值分解法来去除奇异点、改进的张正友标定算法等手段提高了标定精度。实验表明, 在坐标测量精度均为25 μm, 距离测量的系统最大误差小于100 μm, 指标能满足系统要求。

In inertial confinement fusion(ICF) system, manual way has been used to collimate the pellet traditionally, which is time-consuming and low-level automated. A new method based on Binocular Vision is proposed, which can place the prospecting apparatus on the public diagnosis platform to reach relevant engineering target and uses the high precision two dimension calibration board. Iterative method is adopted to satisfy 0.1 pixel for corner extraction precision. Furthermore, SVD decomposition is used to remove the singularity corners and advanced Zhang’s calibration method is applied to promote camera calibration precision. Experiments indicate that the RMS of three dimension coordinate measurement precision is 25 μm, and the max system RMS of distance measurement is better than 100 μm, satisfying the system index requirement.