Research And Implementation Of Automatic Recognition Technology For Gas Meter Based On Linux

With the accelerating pace of urbanization,the urban resident's penetration of natural gas gets higher and higher.Safeguarding the safety of equipment associated with the use of natural gas became a top priority of the quality inspection department.According to national laws,Gas meter as an important device for measuring the amount of gas used,must undergo a mandatory inspection of the quality inspection departments before the user to use.Gas meter as measuring instruments,the accuracy of measurement related to the vital interests of each user,so the metering error is one of most important parameters to be tested.Traditional gas meter error checking method is not only complicated and also low efficiency.The increasing gas meter usage,quality inspection departments testing pressure is growing.This dissertation analyzed the reason of low efficiency to use traditional method,and individually developed and implemented an automatic gas meter recognition algorithm to improve the method based on image processing.At first based on embedded Linux system,this dissertation set up a universal image processing platform,introduced the core processor and camera selection of the platform,and also introduced the establishment process of the image processing application development environment.Secondly,according to the characteristics of mechanical gas meters,this dissertation developed an efficient error-check algorithm.The implementation process of the algorithm was described in detail from three parts: automatic location algorithm,background screening algorithm and identification algorithm.In automatic location algorithm,it made a good positioning precision using color features and frame difference method;In background filter algorithm,this dissertation proposed two species program of background selected according to different flow.It improves the efficiency and accurate rate of errors check;In recognition algorithm,using shift difference method,gradient difference method and direction features difference method,ensured the high reliability of errors check.Thirdly according to the characteristics of electronic gas meters,this dissertation developed an algorithm for electronic LCD screen gas meters specifically,it was also descripted from two difference aspects: automatic positioning algorithm and recognition difference algorithm.In automatic positioning algorithm,it determined the accurate positioning of the character on the LCD screen,using image enhancement algorithm,adapted threshold algorithm and projection location algorithm;in recognition algorithm,it judges character changes directly using before and after frame difference algorithm.This dissertation has tested and used actually to the algorithm and methods.Compared with the traditional method,this method was not only simple to operate,but also more versatile.The efficiency of error checking was significantly improved.