Development Of Automatic Intelligent Verification System For Illuminometer

At present,most domestic illuminometer tests rely on manual operation.Generally,two testers are required to complete the verification process.Not only the entire verification process takes more time,but also because there are more manual operations in the verification process,each part may be intentionally or unintentionally changed by the verification data,so it is easy to introduce the influence of coarse errors and affect the accuracy of the measurement results.The relatively new type of verification instruments mainly rely on the introduction from abroad,and their verification systems are relatively independent,and the degree of automation is not high,which has the disadvantages of low verification efficiency,low verification accuracy,high verification cost and low automation.The future illuminometer verification system must be developed in the direction of automation,intelligence,high precision and credibility.Therefore,this thesis is based on the "Development of automatic intelligent verification system for illuminometer ",researching the visual acquisition technology of low contrast illumination count value in darkroom based on SSD(Single Shot MultiBox Detector),alignment technology of light sensitive components of illuminometer,as well as the automatic control and information processing technology of the verification device of illuminometer,and so on,thus promoting the application and development of the illuminometer verification,improving the accuracy and efficiency of the illuminometer verification,which has important academic value and practical significance.The project research was funded by the Science and Technology Project(2018CJ12)of Guangdong Quality and Technical Supervision Bureau.The paper developed an automatic intelligent verification system for illuminometer,from three aspects——illuminometer intelligent verification device,visual acquisition method of digital instrument value and instrument alignment technology,summarizing the domestic and foreign research progress and then we confirmed the research content.The main work of the thesis includes:? Analyzing the actual requirements of the illuminometer automatic intelligent verification system,designing the illuminometer intelligent verification system verification framework and process,and studying the visual acquisition of the low contrast illuminance count value of the darkroom,the alignment of the illuminometer light sensitive component,and implementation methods of the technology such as the automatic control of the illuminometer verification device as well as the information processing so as to comprehensively improve the level of automated intelligent verification of the illuminometer verification system.? According to the actual environment and the distribution of the positions to be inspected,analyzing the visual acquisition mechanism of the low contrast illumination count value of the dark room based on SSD.According to the verification requirements,improving the SSD target detection process.Studying the industrial camera exposure time and the influence of the pre-processing of the captured image on the subsequent recognition.Designing the illuminometer value recognition comparison experiment.? In-depth analysis of the manual alignment mechanism of the light-sensitive components of the illuminometer,putting forward an automated upgrade scheme,studying on the practical application of the Mask R-CNN deep neural network in the position recognition and alignment of the light-sensitive components,and establishing the functional relationship between the number of pixels and the actual distance,by stepping motors to achieve highprecision alignment of the light sensitive components.? Designing the automatic control software function of the illuminometer verification device and the automatic control and information processing flow of the illuminometer verification device.Completing the design of the automatic intelligent verification system of the illuminometer,and finally entering the information into the report to generate a complete test report.