The Design And Development Of Water Meter Reading Recognition System Based On Image Processing

As a tool for measuring water use,water meters play an important role in urban smart water conservancy construction.To ensure the accuracy of the measurement,the meter needs to be verified before leaving the factory,and can enter the market only after passing the test.In the existing verification methods,many companies still use manual reading measurement and calculation of water meters,heavy workload and low efficiency.Therefore,the water meter reading recognition system with high efficiency,high precision and can provide management function has a high application value for water meter manufacturers.This paper investigates the environment and demand for reading verification by water meter manufacturers,according to the research progress of relevant parties,using image processing and pattern recognition technology,with the core goal of achieving automatic verification of water meter readings,designed and developed a reading recognition system for this water meter.Compared to manual verification,this system can greatly improve the efficiency when multiple water meters are checked at the same time,and ensure the accuracy of identification;Compared to other reading verification systems,this system can simultaneously identify many different types of water meters,solved the problem that the recognition type is more limited,the main research content and results of the paper are as follows:(1)Combine the effects of the hardware part,the paper proposes and uses the idea of template recognition to identify water meters,first of all,collect the template image of the water meter to be tested,then frame and save the recognition areas such as numbers and pointers in the template image.During the verification process,for the acquired water meter image to be identified,according to the position of the identified area in the template image,use Hough transform and related algorithms to accurately locate the area to be recognized in the image to be recognized and use the corresponding recognition method to obtain the result,the verification of different types of water meters can meet the requirements by configuring corresponding template images.(2)By exploring the type of meter and image recognition technology,designed and implemented a development route that uses different image recognition methods in different areas.In the digital area,adopt improved template matching method and Tesseract reading method for pattern recognition.Among them,for half characters,according to the proportion of each character in the half-character image to be recognized,a template half-character is generated to match and the reading result is obtained according to the characteristics of the water meter;all characters are based on template matching,introduce Tesseract for font training and use the training results to recognize all characters with insufficient recognition accuracy in template matching;The pointer area first extracts the pointer,then determine the rotation angle according to the line between the center and top of the pointer to get the result.(3)Using C#,EmguCV,ADB commands and other related technologies,developed a Windows-based form application program.The program incorporates image recognition algorithms,pattern recognition and provides progress review,reading correction,data storage,data export,one-click recognition and other functions are put into the testing phase,laid the foundation for practical use.