The Design Of System Of Automatic Meter Verification And Management Based On Image Processing Technology

With the development of modern information technology, more and more intelligent systems have been used in industrial inspection. Tianjin meter testing station need to test more than a thousand meter everyday. Currently the station is still using the traditional manual test mode, so the workload is heavy and the test efficiency is not stable. Therefore, developing a automatic meter management system with high efficiency and accuracy is very important to Tianjin meter testing station.According to the actual work of meter testing station, a automatic meter management system has been designed in this paper. The system mainly consists of two parts. The first part is the management client, which is used to record and search the employee information, the cooperative enterprise information and the meter information in the station. The second part is the automatic verification system, which is designed to achieve the automatic identification of water meters with the capabilities of acquisition, analysis and processing data. The system radically eliminates the uncertainty brought by the traditional meter reading and improves the efficiency of water meter test.At present, two kinds of meters are being tected mainly in this station, pointer meter and the hybrid type meter combined with digital and pointer. So this issue will identify the two types of meters. Through the selecting of the appropriate image processing and recognition algorithms according to the different dial features, the system achieves the automated verification in this paper.The recognition course of pointer meter is showed as follow. Firstly the dial image should be optimized with image preprocessing algorithm; secondly the large dial, sub-dial and center of pointer should be located according to the priori information, so the possible target area can be separated. Then through the thresholding method constrainted by area conditions, the pointers can be extracted. At last the multi-feature template matching method has been used to recognize pointers’ deflection angle, and according to the relationship between the pointer readings and the deflection angle, the meter value can be gotten.The first step of identification of the digital meter is the segmentation of the display area, using fast two-dimensional Otsu algorithm, which has a good effect on digital extraction. Secondly, create higher degree of specialized digital template which is matching to the testing meter, and the whole number will be recognized with the multi-feature template matching method. And the half word will fill a whole word size with the background according to the baseline of half word. Combined with the height ratio, it can be recognized with the same method used in the whole word.Plenty of image detection experiments and results of value reading on water meter proves several conclusions. The dial locating method combined with priori information can reduces system calculating works, and lower the modification cost. The special templates designed according to the features of actual images in this paper have a higher matching rate, so the accuracy has been improved. Meanwhile, the system has good stability and better adaptability in low-contrast image recognition.