Auto Rack Girders Assembly Holes Measurement Based on Multi-camera Vision

Abstract

Since single camera’s visual field is limited, the measurement method for auto rack girders assembly holes based on multi-group of binocular vision is proposed. The measurement area is divided into several subregions, the measurement data of each subregion is obtained from the binocular vision measurement system, and a larger planar target is used to achieve three-dimensional data registration among adjacent subregion. Since the texture information of truck side-member surface is not abundant, it is difficult to seek the match points on the edge of assembly holes. It is proposed that pasting marked points around the edge of assembly holes for seeking match points. Every two marked points can be connected into one line, and the intersections of the lines and assembly holes’ edge are seen as match points. At last, the geometric parameters of spatial circle are fitted according to its geometrical properties. Experimental results show that the matching difficulty will be avoided effectively, the measurement error caused by perspective projection distortion can be reduced, and the method has higher measurement accuracy.