Research And Implementation Of Detection System For Optical Fiber Quality Parameters Base On Machine Vision

With the wide application of optical fiber in all walks of life,the requirement for fiber is getting higher and higher in the market,especially the special fiber with additional effect.For example,in the medical field,optical fiber is often used to make endoscopic illumination to perform complex surgery.At this time,the quality of the optical fiber will largely affect the condition of the surgery.The rapid and accurate automatic detection of fiber quality parameters in optical fiber grinding process can ensure the quality of optical fiber and improve the production efficiency of optical fiber,which is of great significance.The mode field distribution in the optical fiber is approximately Gaussian when the laser beam emitted by the semiconductor laser is transmitted through the optical fiber.Therefore,the optical fiber divergence angle,deflection angle and other quality parameters can be detected according to the Gaussian beam analysis method.An optical fiber quality parameter detection scheme based on machine vision is proposed in this dissertation,for the characteristics of optical fiber quality parameters.Starting from the requirement of the detection system,the research and design of three key parts of the detection system hardware,image processing and parameter extraction algorithms and detection system software are carried out.The study is as follows:(1)Five-dimensional electronic control platform based on three-dimensional translation stage and two-dimensional angular stage is designed to adjust the angle of incidence and position of the fiber.Based on Programmable Logic Controller(PLC)of Mitsubishi FX3 U series,the positioning control system for electronic control desk is realized.Design the formats and protocols of communication between PLC and computer.(2)For the spot characteristic parameters,according to the detection principle of Gaussian beam,an improved weighted center of gravity method is designed to extract the center coordinates of the spot.The pixel gray value is used as a weighting value to ensure that the center of the spot is closer to the center of the light intensity.For extracting the beam width parameters,a light spot curve fitting method based on Letts criterion method was proposed to calculate the beam width of the spot.Using the statistical principle of the Letts criterion to eliminate the abnormal pixels in the laser spot image and re-fit the spot edge curve to achieve accurate extraction of the laser spot width.Using the statistical principle of the Letts criterion to eliminate the abnormal pixels in the laser spot image and re fitting the edge curve of laser spot to achieve accurate extraction of the laser spot width.After obtaining the spot parameters,this dissertation comes up with the subject of “Similar clustering algorithm” approach based on Cluster Theory which can extract fiber parameters,according to the principle of fiber parameter calculation and the reference clustering theory.After full testing,this method can eliminate human error and increase the measurement accuracy of fiber parameters.(3)The detection system algorithm is programed by combining the OpenCV vision library with the C++ language under the Windows platform.The upper software is developed by using MFC and a set of automated fiber optic parameter detection system is designed.Finally,the performance and accuracy of the system are analyzed and tested,and the applicability of the system is analyzed by the GR&R method.The test results show that the detection error percentage of the detection system designed in this dissertation is less than 2%,which meets the requirements of the detection system.