Testing System Based On Image Recognition Research And Development Of Water Meter Flow Performance

The accuracy of the water meter is related to the interests of the majority of users.The state stipulates that the water meter needs quality inspection before leaving the factory.At present,some domestic powerful water meter enterprises have realized the high efficiency and low cost of water meter automatic verification method.However,the optimization and improvement of automatic verification system still has a large space.In this thesis,a set of performance evaluation system for the internal mechanism of water meter is designed based on image recognition.It mainly includes the verification of water meter and the performance analysis of internal mechanism.This system can be applied to the development,production and improvement of water meter.The design ideas of this system are as follows:Firstly,according to the requirements of the performance test system,the construction scheme of the overall hardware platform is designed.For the identification part of the water meter dial information,the selection of the visual device is completed.At the same time,the corresponding lighting scheme is designed for the image acquisition of the water meter dial.Finally,the hardware platform of the water meter test is completed.Then,different detection schemes are designed according to the water meter test under the flow points of high flow area and low flow area during the water meter test.Among them,for the water meter test in the case of high flow area,the measurement unit identified by the system is the last pointer.The recognition process of the last pointer is as follows:firstly,Hough circle detection is used to locate the dial and the area where the last pointer is located.The center of the circle in the circular area where the last pointer is located is used as the rotation axis.Then,the last pointer and contour are extracted by HSV color space and Canny operator.Next,the deflection slope of the last pointer is calculated by improving the circumferential gray detection method and converted into the corresponding deflection angle.Then,the deflection angle of the last pointer of each adjacent frame is subtracted to obtain the rotation angle of the last pointer at each moment.For the water meter test in the case of low flow area,the measurement unit identified by the system is the sensitive pointer.The recognition process of the sensitive pointer is as follows: firstly,the YOLOv5 training detection model is used to locate the position of the sensitive pointer.Then,the SIFT feature matching algorithm is used to establish the relationship between the adjacent frame sensitive pointer images.Next,each pair of matched feature points is converted into polar coordinates and the sensitive pointer rotation angle data set is calculated.Then,the RANSAC algorithm is used to eliminate the mismatching points and obtain the correct sensitive pointer rotation angle.The cumulative flow and rotation acceleration of the process are calculated by the rotation angle and frame rate of the target pointer in each adjacent frame image.At the same time,the rotation acceleration data is visualized.According to the evaluation criteria of the system,judge whether the water meter is qualified.And quantitative evaluation of the internal mechanism of the water meter.The feasibility of the system is proved by field experiments.The design of this system provides a new method for water meter enterprises to develop and produce water meters.