超大视场红外凝视成像系统内外参数的标定与评价

Calibration and Evaluation of Internal and External Parameters of Super-Wide Field-of-View Infrared Gaze Imaging System

针对现有成像系统标定方法无法有效校正超大视场红外凝视成像系统的问题,提出了一种新的间接角点检测算法和针对Scaramuzza校正模型中迭代计算会出现局部最优问题的改进方案。利用形态学操作对棋盘标定板边缘进行像素级检测,再使用插值技术细化边缘,使边缘具有亚像素级精度,获取真实角点附近的4个棋盘单元角点,对4个角点进行坐标平均化间接获取真实角点坐标,使角点检测正确率达到100%,远高于一般算法,并且更加贴合真实角点位置。在逐次迭代过程中,通过对整个区域使用特定圆形区域采样网格点采集SSRE,再将所有采集数据进行最小化取值,既避免了陷入局部最优的情况,提高了定位精度,又大大减少了迭代次数和迭代运算量,弥补了现有一般标定算法无法满足超大视场红外凝视成像系统的不足。

Based on the fact that the existing calibration methods for imaging systems cannot effectively calibrate the super wide field-of-view infrared gaze imaging system,this research proposes a new indirect corner detection algorithm as well as an improved scheme to address the problem of local optimization in the iterative computation of the Scaramuzza calibration model.With the help of morphological operations,pixel-level detection is performed on the edges of the chessboard calibration board.Interpolation techniques are then applied to refine the edges,achieving sub-pixel accuracy.This allows for the acquisition of four chessboard unit corners near the true corner points.By averaging the coordinates of these four corners,the coordinates of the true corner points are indirectly obtained,resulting in a corner detection accuracy rate of 100%,which is significantly higher than that of general algorithms and more closely aligns with the true corner positions.During the iterative process,the entire region is sampled with a specific circular region sampling grid to collect the Sum of Squared Residuals Errors(SSRE).By minimizing all collected data,this approach not only avoids falling into local minima,improving positioning accuracy,but also significantly reduces the number of iterations and computational load.It effectively addresses the shortcomings of existing general calibration algorithms that cannot meet the requirements of super-wide field of view infrared gaze imaging systems.