Online defect detection method of optical cable pitch based on machine vision technology and deep learning algorithms

In the field of optical cable manufacturing, the pitch error of finished optical cable exceeding 2 % is considered a defect. Currently, there is no method for online detection of pitch defect in optical cable. Therefore, this paper proposes an online intelligent detection method for cable pitch defects. This method utilizes a multi-sensor fusion image acquisition platform consisting of a Doppler laser velocimeter and an industrial line scan CCD to achieve visual information acquisition of ultra long measured objects. A mathematical model that integrates cascaded dual deep convolutional networks and iterative approximation algorithms based on deep learning technology is proposed for defect detection of optical cable pitch. In order to verify the accuracy and stability of the on-line detection method, 6 types of optical cables were prepared for experimental testing. The experimental results show that the method is robust in the field of engineering detection, and the detection results meet the accuracy requirements.

Automated summary

This paper proposes an online intelligent detection method for cable pitch defects in the field of optical cable manufacturing. The method utilizes a multi-sensor fusion image acquisition platform and deep learning technology to achieve accurate detection of pitch defects in optical cables.