深度相机的测量误差建模及校正

Modeling and correction of measurement errors based on depth cameras

ToF(Time of Flight)深度相机是获取三维点云数据的重要手段之一,但ToF深度相机受到自身硬件和外部环境的限制,其测量数据存在一定的误差。本文针对ToF深度相机的非系统误差进行研究,通过实验验证了被测目标的颜色、距离和相对运动等因素均会对深度相机获取的数据产生影响,且影响均不相同。本文提出了一种新的测量误差模型对颜色和距离产生的误差进行校正,对于相对运动产生的误差,建立了三维运动模糊函数进行恢复,通过对所建立的校正模型进行数值分析,距离和颜色的残余误差小于4 mm,相对运动所带来的误差小于0.7 mm。本文所做工作改善了ToF深度相机的测量数据的质量,为开展三维点云重建等工作提供了更精准的数据支持。

Time of Flight(ToF)depth camera is one of the important means to obtain three-dimensional point cloud data,but ToF depth camera is limited by its own hardware and external environment,and its measurement data has certain errors.Aiming at the unsystematic error of ToF depth camera,this paper experimentally verifies that the color,distance,and relative motion of the measured target affect the data obtained by the depth camera,and the error effects are different.A new measurement error model is proposed to correct the error caused by color and distance.For the error caused by relative motion,a three-dimensional motion blur function is established to recover it.Through the numerical analysis of the established calibration model,the residual error of distance and color is less than 4 mm,and the error caused by relative motion is less than 0.7 mm.The work done in this paper improves the quality of the measurement data of the ToF depth camera,and provides more accurate data support for 3D point cloud reconstruction and other work.