基于条纹反射法的平面镜面形测量

为了评价平面镜面形误差,采用基于条纹反射法的面形检测原理、通过平面镜法线偏移实现面形误差评价的方法。将正弦条纹投射至光屏,经被测平面镜反射至相机,通过小孔成像原理和四步相移法获得被测平面镜上一点的法线,并与经最小二乘算法面形重建的该点理论法线比较,获得该点法线偏移,利用法线偏移实现对平面镜的面形评价。对测量方法进行计算机仿真,以证明其正确性;搭建实验测量系统,并对某一款手机屏幕进行测量实验。结果表明,该测量方法的重复性精度优于1mrad,测量重复性高。该研究为玻璃面板等具有高反射特性的平面面形误差提供了参考。     

多尺度特征融合的目标检测算法研究

针对Faster R-CNN在多尺度目标检测时易出现小目标漏检和误检的问题,提出一种改进的多尺度目标检测算法。将利于小目标检测的低层网络和利于大尺度目标检测的高层网络进行多尺度特征融合;在训练阶段,采用在线难例样本挖掘算法维护难例样本分类池,加速神经网络模型迭代收敛,解决训练样本不均衡、训练效率低下的问题;计算并统计待检测目标的尺度大小,合理控制用于生成候选区域的锚框尺寸,提高模型泛化能力。采用PASCAL VOC2012公开数据集和类人足球机器人自建数据集进行算法验证,实验结果表明,相比Faster R-CNN算法,本算法的平均检测精度在上述数据集下分别提高了8.61和5.47个百分点。     

摄影测量相机成像模型参数的单图标定方法

为研究热控措施下温度波动引起的相机成像模型参数变化规律问题，提出了一种摄影测量相机成像模型参数的单图标定方法。摆放多张不同位置及姿态的同规格同精度标定板于相机视场内，为相机单图标定提供大量不同深度不同姿态及不同位置的已知空间点阵约束信息，在有限空间内的连续时刻分别采集单张标定场景图像，通过多站位光束平差算法实现连续时刻下相机成像模型参数的标定。实验结果表明，此标定方法的空间点像面反投影误差标准差为0.48μm,标定精度为1/15像素，可为后续变温条件下相机成像模型参数变化规律的探究提供准确的温度与相机参数相对应的数据集。     

工业相机自热引起像点漂移模型与补偿方法

In order to reduce the influence of temperature on the image point coordinates of industrial cameras in visual measurement, the image point drift caused by the self-heating of the camera was studied, and a compensation method for the thermal image point drift of industrial cameras was proposed. The finite element simulation analysis of the industrial camera model through Ansys Workbench shows that the self-heating of the industrial camera will cause the imaging optical path change and sensor expansion change, quantitatively the influence of the optical path change and sensor expansion on the image point coordinates was analyzed, and the image point drift compensation model was established. A large number of experiments have shown that the image point drift error compensated by the model is reduced from 0.4-0.6 pixel to 0.1-0.2 pixel, which is equivalent to the image point drift suppression effect achieved by hardware thermal control. However, compared with the thermal control device, the method of using the model for compensation has obvious advantages of simple structure and low cost. The temperature compensation model proposed in this research provides a theoretical basis for reducing the image point drift error caused by the self-heating of the camera in the visual measurement.     

最小化重投影误差的手眼标定优化算法

为了减小图像噪声对手眼标定精度的影响,在传统的手眼标定算法的基础上,利用基于最小化重投影误差优化算法求解手眼标定方程.以基于矩阵直积参数化方法计算出的手眼关系矩阵为初值,以摄影测量光束平差为优化模型,最小化重投影误差为代价函数,获得最优的手眼关系标定值.实验结果证明了该方法的具有较好的稳定性和精确性,算法求解了 10次...