本文介绍了一种移动式三维视觉测量系统,用于大型物体自由曲面的测量。传感器采用双目立体视觉方式,并结合编码结构光技术,使匹配过程简化。传感器移动到大型物体周围对各局部区域进行测量。采用一个平面靶标作为中介,建立了传感器位姿变换矩阵的优化目标函数,根据靶标上多个特征点求解出该矩阵,即拼接矩阵。传感器在各个位置所测三维数据通过拼接矩阵统一到全局坐标系下。视觉测量系统在2个位置对维纳斯石膏像进行了测量,并进行拼接。结果表明,该测量系统操作简单,适用范围广;x、y、z坐标拼接RMS误差分别为0.038mm、0.022mm和0.135mm。
A mobile 3D vision measurement system for large-scale free form surtace is introduced. The sensor of the system adopts binocular stereo vision principle combined with coded structured light technology, which simplifies the process of finding correspondence. The sensor is moved around a large-scale object, and each local surface is measured. Using a planar target as intermedium, an optimized object function of the position and orientation transformation matrix of the sensor is established; and the matrix, namely registration matrix, is derived from the feature points on the planar target. The 3D data obtained at the local locations are unified in the global coordinate frame using the registration matrix. This measurement system was used to measure a Venus plaster model at two local locations, and the 3D data were registered. The experimental results show that the measurement system features simple operation and wide application fields; the registration RMS errors for x,y and z coordinates are 0. 038 mm,0. 022 mm and 0. 135 mm respectively.