A box detection system and method based on binocular vision

Abstract

The present invention provides a box detection system based on binocular vision, which includes: several first measuring tools, which are respectively arranged on the partition, the first web, the second web, and the bottom plate inside the box. There are a number of marking points on the non-uniform outer surface of the section; several second measuring tools are fixed on the outer wall of the first / second web of the box body, and a marking point is set on the top of the second measuring tooling; two industrial The camera is used to collect images of all first measuring tooling mark points in the measuring unit from above the box, and collect images of all second measuring tooling mark points on the first / second web; the signal processing unit, according to all the images in the measuring unit The image of the first measurement tool, detect the distance between the partitions, the verticality of the partition, the verticality of the first / second web, and generate the straightness of the first / second web and the root of the bottom plate according to the image of the marking points of the second measurement tool test results. The invention also includes a box detection method based on binocular vision.

Description

technical field [0001] The invention relates to the field of mechanical manufacturing, in particular to a box detection system and method based on binocular vision. Background technique [0002] Large-sized boxes are the most critical components in the construction of large-scale lifting machinery, marine engineering equipment, ships, etc. The quality and performance of their manufacturing directly affect the safety and use of the equipment, and their manufacturing efficiency also directly affects the manufacturing cycle of the equipment. . As we all know, the structure of box girder is complex, with many partitions, large size, and narrow space, which is prone to welding deformation. The assembly accuracy of components before welding and the monitoring of deformation after welding are particularly critical in the manufacturing process of box girder. At present, the pre-weld assembly accuracy detection of large-scale cabinets in my country basically adopts manual measuremen...

AI Technical Summary

Problems solved by technology

At present, the pre-weld assembly accuracy detection of large-scale cabinets in my country basically adopts manual measurement methods, which have low detection efficiency, high environmental pressure, and high labor intensity for workers. The measurement results are not objective enough and the integration of measurement information is not high.

[0003] Since the three-dimensional coordinate measurement of large box girder components has the characteristics of large measurement range, high precision and real-time measurement on site, it is required that the measurement technology and equipment must meet the conditions of accuracy and environmental adaptability at the same time. Traditional measurement methods mostly use manual ruler measurement and Theodolite measurement method has a long measurement period and consumes a lot of labor, and the measurement data cannot be fed back in real time, which often leads to quality problems such as unqualified assembly accuracy and large welding deformation, resulting in a large number of rework and repair work

At the same time, manual measurement methods cannot truly meet the technical requirements of automated production for high-precision, large-scale 3D overall on-site testing

[0004] With the development of manufacturing process automation and manufacturing data informatization, the current backward measurement methods have become the bottleneck of common detection technology in the manufacturing and production of large and complex structural parts

Images