Research And Implementation Of Preform Inspection Detection System Based On Machine Vision

As the cornerstone of the communication industry,optical fiber is the most fundamental material foundation for the development of the communication industry in recent communication field.Many communication manufacturer have captured the business opportunities in the optical fiber manufacturing industry and launched the business of optical fiber production.The quality of the preform directly determines the quality of the produced fiber because of being the core raw material of optic fiber.The fiber production would be difficult if preform contained too many impurities,and the produced fiber would be ill-quality.So the detection of preforms play an essential part of the process of manufacturing optical fibers.At present the detection of optical fiber preforms almost relies on the naked-eye detection.The advantages of naked-eye detection are low technical request and wide application environment.However,the fatal flaw of this detection is the low accuracy and detection efficiency which is restricted by the requirements of the working environment and the work condition.To resolve defects this thesis designs an automatic detection system based on machine vision,which extracts the enamel by means of image processing to realize the detection of preforms.The system is fast and accurate,and has the advantages of ease of use and maintenance.The main work and innovations of this thesis are:(1)The detection of flaws in prefabricated rods is generally artificial naked eye detection at present.In this thesis,machine vision technology is used instead of human eyes to obtain information,and image processing technology is used instead of human brain to complete the detection process.A complete system for preform detection is designed and implemented.(2)The defect detection image is deeply studied,and the image preprocessing and flaw detection processing are realized.In the image processing method,the point-bypoint method for image block processing is selected,the image is divided into blocks,and the threshold is calculated from the first row of pixels to calculate the threshold point by point,which avoids the globalization of the threshold.At the same time,the point-by-point method was adopted and targeted optimization was carried out for the applicable project.According to the particularity of the unilateral lighting of the project,the threshold matrix obtained by the point-by-point method was further processed,adding a parameter characterizing the attenuation of light intensity further optimizes the threshold matrix.This method takes into account the characteristics of the specific image of the preform image and the processing requirements,making the processing result reliable and efficient.(3)The design and implementation of the chirped detection system are studied and implemented from two directions of hardware and software design.The hardware includes light source,lighting system processing and other aspects.The software includes system detection process design and interactive interface design implementation.The system was tested and analyzed from the aspects of the detection efficiency and stability of defects.The results show that the system designed in this thesis can operate stably.Compared with the traditional threshold segmentation detection method,the detection is efficient,stable and meets the design requirements.