Research On Foreign Body Detection Method And Application Of Equipment Surface Based On Structured Light

Foreign body detection on equipment surface involves many fields,such as aerospace,high-speed train,industrial equipment and so on.This paper takes high-speed train as the research object to conduct research.The detection of foreign body on the surface of high-speed train is a powerful guarantee for the safe operation of the train.In order to avoid the falling off of foreign body on the surface of the high-speed train during operation and causing safety accidents,it is necessary to detect and deal with foreign matters regularly.The traditional foreign body detection mainly relies on manual,the detection efficiency is low and the accuracy cannot be guaranteed.Therefore,it is urgent to explore an intelligent detection technology for foreign body on the surface of equipment to reduce human investment and improve the detection efficiency and accuracy.With the rapid development of camera,digital projector,computer and other hardware equipment as well as image processing technology,machine vision has become a research hotspot in the field of intelligent detection,which provides a new possibility for foreign body detection on equipment surface.Machine vision is a technology that collects images by camera and processes images to obtain target information.However,due to the increasingly complex practical application scenes,only relying on two-dimensional image information is insufficient to achieve accurate discrimination.Therefore,machine vision technology is gradually transformed to three-dimensional.Among them,3D measurement technology based on digital fringe projection has been widely used in industrial testing,quality control and other fields because of its advantages such as non-contact,fast measurement speed,high measurement accuracy and strong robustness.In this paper,three key problems of phase extraction and expansion,phase error compensation and system parameter calibration in the 3D measurement technology based on digital fringe projection are analyzed and optimized.The proposed method was applied to the field of foreign body detection on equipment surface and its feasibility was verified by experiments.The specific research contents are as follows:First of all,accurate phase extraction and unwrapping is the premise to ensure the accuracy of 3D measurement technology based on structured light.Theoretical analysis and experimental research were carried out on the phase extraction and unwrapping algorithm,and the corresponding solutions are proposed for the following problems:(1)An effective phase extraction method based on the standard deviation of light intensity is proposed to solve the problem that the background noise or shadow area in the acquired image lead to invalid area in the relative phase distribution map;(2)A phase error correction method based on Gaussian filter is proposed for the jump error at the periodic junction caused by the fact that the relative phase and fringe orders could not jump simultaneously;(3)To solve the problem of code word demodulation error in the phase-coding method,the method of introducing ?/k phase shift in the coding phase is proposed.The above methods are verified by experiments,and the results show that the proposed methods are simple and effective.Secondly,the phase error caused by gamma distortion of the projector is the main source of the system error.Double N step phase shift method and Hilbert transform method are used to carry out phase error compensation experiments.The results show that the phase error compensated by the two methods mentioned above can be controlled below 0.06rad,which shows a good compensation effect.Then,the system parameter calibration is the key to realize the conversion from 2D image information to 3D space information.Based on the principle of camera imaging,the ideal and distortion system models are established respectively,and the nonlinear mapping relationship between phase difference and three-dimensional coordinates is deduced.The nonlinear mapping relationship between phase difference and three-dimensional coordinates is fitted by the Extreme Learning Machine(ELM)network with excellent Learning speed and generalization performance,and the performance of the network was tested with training samples and test samples respectively.The appropriate number of hidden layer neurons l and activation function g(x)are determined,and the network parameters are saved to realize the calibration of system parameters.The results of 3D reconstruction of training samples and test samples show that the system parameter calibration method based on ELM network has high measurement speed and high measurement accuracy.Finally,the 3D measurement system based on structured light is built,and the foreign body detection experiment is carried out with the underframe cross beam of high-speed train.On the basis of the above research,the optimized algorithm is applied to the 3D reconstruction of the high-speed train underframe cross beam with foreign bodies,so as to realize foreign body detection.The experimental results show that the proposed scheme has high feasibility and measurement accuracy,which plays an important role in promoting the rapid,high-precision and automatic detection of foreign bodies on the surface of equipment.