Detection Of Defects In Nameplate Printing Of Smart Meter Based On Machine Vision

Wit h the cont inuous accelerat ion o f do mestic urbanizat ion,relevant power facilit ies o f the nat ional grid system has been rapidly developed.As a key part of the nat ional grid system,smart meter’s production scale has been also great ly developed.In the manufacturing process of smart meters,in order to ensure the relevant nat ional standards and qualit y requirements o f the smart meters,it is necessary to strict ly test the display qualit y o f the LCD screen and the qualit y o f the nameplate print ing.The meter nameplate is a key part o f the smart meter.The appearance defects of the smart meter nameplate will affect the overall appearance o f smart meter,so the qualit y assurance o f the meter nameplate is very important for the product ion qualit y of smart meters.The main appearance defects of smart meter nameplate are print ing defects such as missed printing,lack o f print ing,ink bleed,and fouling.And according to the actual product ion requirements,the detection accuracy o f print ing defects o f the smart meter nameplate is high.Current ly,the detect ion of nameplate print ing defects o f smart meters is mainly done manually.Considering the high labor intensit y o f manual detect ion,the detect ion qualit y is difficult to maintain stabilit y,and the product io n qualit y o f smart meters can not be guaranteed.Therefore,the auto mat ic product io n line urgent ly needs a set o f system to detect print ing defect s o f smart meter nameplate.This paper takes a certain t ype of smart meter nameplate as the research object,uses image processing methods to detect print ing defects o f meter nameplate,and designs a smart meter nameplate print ing defect s detect ion system based on machine visio n.The main research contents are as fo llo ws:First,build so ftware and hardware framework o f smart meter nameplate print ing defect detect ion system.In the hardware construct ion,the hardware framework is formed by select ing and jo int ing the hardware of the detect ion system;in the so ftware framework,studying C# language writ ing method of software interface and using C#language to write the detect ion system interface.Studying connect ivit y issues o f Halcon machine visio n so ftware and C# language to ensure the operabilit y of the detect ion system.Second,studying extract ion methods o f detectio n areas o f smart meter nameplate.According to the characterist ics of smart meter nameplate structure,detect ion areas o f smart meter nameplate need to be divided int o three parts: bar code area,parameter descript ion area and meter number area.First,using image filtering and image segmentat ion to extract entire detection area of meter nameplate;then use double thresho ld segmentat ion and image expansio n with corrosion to extract barcode area and meter number area;use double thresho ld segmentat ion and manual segmentat io n to extract parameter descript ion area.Through accurate extract ion o f three detect io n areas to achieve co mplete extract ion of meter nameplate detect ion areas.Finally,studying detection methods for nameplate print ing defects o f smart meters.Use three different image processing methods to detect defects in three detect ion areas: using improved template comparison method to detect defects in parameter descript ion area;us ing improved template co mparison method and character recognit io n method to detect defects in meter number area;using method o f adjust ing corrosion and expansio n structural elements co mbined with barcode recognit io n method to detect defects and identify o f barcode area.