Multi-camera array-based large-size industrial plate online measurement system and method

Abstract

The invention discloses a multi-camera array-based large-size industrial plate online measurement method. The method comprises the following steps: S1, constructing an embedded camera array accordingto the shape of an industrial plate and a target size; S2, an ARM is adopted to control a common monocular camera to collect a local high-resolution image of the plate; S3, realizing feature extraction and local measurement of the local high-resolution image target by utilizing an embedded technology; And S4, image splicing is carried out according to the target edge features of the local images,a global image is reconstructed, and the geometric dimension of the industrial plate and accurate segmentation of multiple targets are calculated. According to the invention, a distributed processingarchitecture is adopted, the subsystems synchronously process, the scale of the system has expandability and is not limited by the shape and the target size of an industrial plate, and the measurementprecision and the operation efficiency are high.

Description

technical field [0001] The invention relates to the industrial intelligent application field of machine vision, in particular to a large-size on-line measurement system and method for industrial plates based on a multi-camera array. Background technique [0002] Machine vision technology uses cameras, which can not only achieve comprehensive measurement and detection of parameters and characteristics such as object outline size, area, spatial position, shape, defects, etc., but also have the advantages of non-contact, high efficiency, reliability, good versatility, and easy automation. . In high-volume, continuous automated production lines, there is an urgent need to use vision to achieve automatic identification and accurate segmentation of product targets. [0003] At present, the visual measurement technology for small and medium-sized mechanical devices (below 100mm) is relatively mature, but there are many problems in the measurement of large-sized devices larger than...

AI Technical Summary

Problems solved by technology

[0004] 1) Most visual measurement equipment has a specific range, and cannot measure devices whose size exceeds the range;

[0005] 2) The field of view of the camera is limited, and the monocular camera distorts the edge of the remote target seriously, and there is often false edge information, which seriously affects the accurate segmentation of the target;

[0006] 3) The shape and size of the measurement object are limited by the working surface of the measurement equipment;

[0007] 4) Most of the measuring equipment is relatively independent, and it is difficult to fully integrate with the production line;

[0008] 5) Large-scale equipment is relatively heavy, with weak handling performance and high equipment cost

However, there are inevitably errors in each link of image stitching, and these errors affect the stitching accuracy to varying degrees, thereby affecting the accuracy of device size measurement

[0011] Second, it is necessary to overcome the problem of real-time calculation of a large number of sequence image processing, making online measurement possible

[0013] 1) The field of view of the camera is not enough, and the field of view of the monocular camera is difficult to cover the entire plate plane;

[0014] 2) There are often serious distortions in the acquired image, and there are false edges of the target;

[0016] 4) Measurement and cutting algorithms are highly complex

[0017] Although many scholars have conducted research on these problems and challenges, there is still no suitable online measurement and multi-target extraction of large-size industrial plates based on machine vision in the market

Images