Research On Automatic Visual Verification System For Water Meter

As an instrument for measuring water consumption,water meters are one of the most widely used meters in the field of flow detection.At present,water meter is the instrument with the lowest degree of automatic verification among “three kinds of instruments” used in family.There are some problems with the automatic verification of the water meter such as excessive types of water meter types,inconsistent shapes,and difficulty in positioning the verification position in the dial.Based on the existing verification device,combined with the working principle of water meter,and using image processing algorithm,this paper develops and designs a water meter automatic verification system based on machine vision technology,which makes large-scale automatic verification of water meter possible.The main research contents include:1.Design and build a fully automatic water meter verification device and write verification control software.On the one hand,in the existing laser sensor-based water meter verification device,the water meter fixture,the posture adjustment station and the multi-water meter linkage clamping mechanism were designed,the movable laser sensor mechanism was added,the machine vision technology was introduced,and the robot was automatically connected up and down.Finally completed the construction of the water meter verification device.On the other hand,this paper designs and develops a water meter automatic verification control process and water meter information management program that meets the water meter verification standards,including the flow control of the hardware verification device and the use of the SQLite database to establish the water meter storage water meter verification result.2.The three-frame difference algorithm is used to dynamically extract the position of the plum blossom rotors of different water meters.According to the difference of dials of different water meters,the three-frame difference method and the background subtraction method based on hybrid Gaussian background modeling are used to dynamically detect and extract the rotorarea of the plum.The comparison test results show that the three-frame difference method can not only realize the real-time extraction of the plum rotor,but also achieve effective extraction under the bubble interference.3.Use CSS corner detection and convex hull extraction algorithm to achieve precise positioning of the laser probe sampling points.The Canny detection operator is used to extract the edge of the plum rotor,and the CSS algorithm is used to extract the edge points of the plum rotor.The convex points are extracted and the coordinates of the center of gravity are calculated.According to the geometric characteristics of the plum rotor,the polar coordinate system with the center of gravity as the pole is established.Combine the circle parameter equation formula to calculate the sampling point coordinates.The coordinates of the sample points obtained by the algorithm are compared with the ideal coordinates obtained by the image accurate measurement software.The results show that the error of the localization algorithm is within 1.91 pixels.Repeated positioning experiments on multiple sets of water meters show that the laser probe positioning repeat error does not exceed 0.37 mm.Through the actual verification of the prototype system operation and error experiment,the results show that the error with the manual test results does not exceed 0.54%,indicating that the verification system meets the requirements of the verification procedures,and the automatic verification of the water meter can be realized.