The Machine Vision Inspection System Of Electric Energy Meter LCD Screen Based On LSD

There is a problem of high error detection rate due to the inaccurate positioning of LCD in the automatic detection of the quality of the LCD screen of the intelligent electricity meter.In this thesis,a vision detection system for LCD machine based on LSD(a Line Segment Detector)location method is designed to solve the above problem.The thesis takes the image of the actual electricity meter in the industrial field as the research object,designs and develops the quality detection and processing algorithm for the LCD panel of the electricity meter,bar code recognition module and digital character OCR recognition module.The system based on LSD straight line detection firstly gives the LCD panel precise positioning and tilt correction,realizes the recognition of the appearance of characters on the LCD screen through using the combination of global binarization and local binarization,and then using template matching to judge whether there is anything undectcted.The system finally realizes the function of precisely positioning the LCD screen of the electricity meter,detecting the leak of LCD screen character,identifying the barcode and the information the electricity meter carrying,such as the bottom of the electricity meter.A large number of experimental results show that the system can effectively reduce the error detection rate of the quality detection of the liquid crystal screen of the electricity meter and realize the all-aspect detection of the liquid crystal screen of the electricity meter.The specific work of this thesis is as follows:(1)The machine vision inspection system of the electric energy meter LCD screen based on LSD is designed.including the hardware and software environment of the electricity meter image acquisition.In the aspect of software,including the function of each module,the mode of realization and the interface of the system.(2)In image preprocessing,according to different brightness meter image binarization can not achieve satisfactory results through a single method,this thesis proposes a algorithm by detecting brightness of image to use a variety of binarization methods for image binarization,and implements a variety of brightness image binarization can achieve the result of subsequent processing requirements.(3)In barcode recognition module,according to the rectangle feature of barcode area,this thesis propose a algorithm based on image morphology processing and boundary finding to locate barcode area of electric energy meter.On the basis of the accurate location of the bar code area,the open source barcode recognition toolkit is used to identify the bar code,and the recognition effect is better.(4)In the LCD screen character detection module,this paper studied the characteristics of LCD screen characters,and proposed template matching algorithm using location information and display information as templates.At the same time,in the image tilt correction link,this paper elaborated the inclination angle determination,rotation correction and interpolation algorithm contrast selection and other aspects in detail.On the basis of image tilt correction,the precise positioning of the LCD screen is carried out by the LSD algorithm with the characteristics of the LCD screen boundary.(5)In the digital character OCR recognition module,this paper studied the display characteristics of the digital characters.Based on the location of the LCD,this paper used the way of detecting the number of bright dots of each digit tube to determine the display content of the digital characters.(6)Through experiments on a large number of sample images provided by SanHui company,the standard template is perfected,and the algorithm of locating LCD and the algorithm of character detection are optimized.As the result of testing 5000 unidentified data provided by Zhengzhou Sanhui electrical Company Limited,the recognition accuracy of the algorithm is over 90%.Experiment results show that the recognition algorithm proposed in this thesis has more satisfactory accuracy.