Multi-Camera Measurement System Based on Multi-Constraint Fusion Algorithm

In order to improve three-dimensional measurement accuracy in the large field and to overcome the shortcomings of the traditional bundle adjustment algorithm for image correction being not ideal for large field of view(the control points concentrated in the center of the view),this paper presented a multi-camera 3D measurement methods based on multi-constraint fusion algorithm.The basic idea is collinear equation of three-dimensional space coordinates of the control points and corresponding pixel coordinates is looked as constraints,using the known distance constraint,three-point collinear,four-point coplanar to establish the restriction between coordinates of the control points and pixel points,thus completing the three-dimensional coordinate measurement,at the same time,achieving the online calibration of the system parameters.In the experiment,the RMS error and relative accuracy of distance are used to analyses the measurement accuracy,the relative accuracy of distance reaches to 1∶7 000~1∶15 000,compared to the traditional bundle adjustment algorithm,the measurement accuracy of this algorithm is improved an order of magnitude,it is a reliable high-precision vision measurement method.