Design And Implementation Of Self Service Gas Meter Reading System Based On Image Recognition

Recently years,with the improvement of the urbanization level in China,the total amount of natural gas is increasing,which increases the difficulty of gas meter reading.Gas meter is an instrument to measure the amount of natural gas.There are many types of gas meters in the market,among which membrane gas meter accounts for a large proportion.Reading membrane gas meter requires traditional manual way which need to get in users’ houses.This way is not only inefficient,but also has certain risks when reading meters in special circumstances.Therefore,it is pain and difficult to use this kind method for gas meter reading and charging.The existing remote meter reading schemes based on machine vision depend on adding auxiliary photo and upload devices to the gas meter terminal.But this scheme is not only expensive but also non-universal and even difficult to be implemented in a large scale.Under the premise of not changing the original structure of the gas meter,this thesis designs a simple and versatile semi-automatic meter reading scheme.The gas user takes pictures of the gas meter and uploads them to the cloud host.In the cloud host,the image recognition technology is used to extract the user’s gas meter number and gas consumption.The main research of this thesis includes the following aspects:1.The technology of region of interest(ROI)extraction and character segmentation is studied.The counter area and number area of gas meter are regarded as ROI.The digital characters of ROI should be extracted.Firstly,image preprocessing technology is used to improve the image quality.For counter area of gas meter,the color characteristics of this area are analyzed,and a further image processing method using image color saturation component is proposed to obtain the image which can extract the counter area.For the number area of gas meter,the binary image containing only bar code is obtained by using image gradient and morphological operation.Finally,the digital characters of ROI are segmented by the projection.2.The digital character recognition technology of gas meter is studied.Firstly,the characters of the counter area are divided into complete characters and incomplete characters.Aiming at the problem of low recognition accuracy of incomplete characters,the method of making incomplete character samples is adopted to enrich the samples of digital characters.Then,selecting the number recognition algorithm and set the parameters of the algorithm.Finally,the recognition algorithm with high accuracy of gas meter digital character recognition is selected as the algorithm of gas meter digital character recognition.3.Combined with the above research,a gas meter reading system based on image acquisition and recognition is designed and developed.The experimental results show that the real-time performance and accuracy of recognition meet the requirements of practical application.