Research On Digital Image Recognition Of Gas Meter Readings

With the rapid development of national economy and further promotion of urbanization construction, the number of users of urban residents gas continues to increase, this increasing the gas meter reading and billing settlement workload. Existing handheld meter reading machine and IC card prepaid mode, has not been meet the need of convenient clearing gas meter. Although there have been many remote meter reading system for different applications on the market today, to a certain extent, solve the difficult problem of meter reading. But the meter reading system to the original gas meter need to tear open outfit, add some electronic devices, the performance of the original gas meter will cause certain influence. Under article without changing the structure of the gas meter premise, combined with previous topics, this paper based on digital image processing technology to identify gas meter reading system. The main research contents of this paper are as follows:In the image pre-processing stage, because the noise will inevitably be a certain degree of damage and the variety of the transmission and storage of the image, resulting in lost the essence or deviate from the needs of the people, and this requires the acquisition of image enhancement. Through the comparison of various image enhancement technology, found the gray correction, image smoothing and de-noising in image processing has a very good effect; comparison of several different filtering methods, the image is filtered using median filtering method. In addition, according to the Otsu algorithm can automatically obtain the best image quality threshold increases the visibility of the image, so using Otsu algorithm for two value image after filtering.In this paper, one of the most important and difficult step is number area location and segmentation for the system design. The accuracy of that directly affects the accuracy rate of single characters segmentation and number recognition. The paper by the method of horizontal and vertical projection positioned to achieve the region’s gas meter digital coarse positioning. Then the Canny operator edge detection and mathematical morphology algorithm which combines the re-positioning image of coarse positioning, to achieve precise positioning of the gas meter reading area. As the captured images inevitably suffer from different degrees of tilt, we use the Canny edge detection and Hough transform to correct the tilting images.In the area of digital reading recognition stage, using the BP neural network to recognize the reading. In the test phase of the BP neural network, through a number of experiments to determine the most conducive to achieve the desired error and the expected output of the hidden layer neurons number. The training algorithm of BP neural network to train the network, to achieve the best training algorithm is obtained by comparing the training effect and the highest recognition rate, as the final algorithm of the design. Finally, through the MATLAB software to recognize the segmented characters, the results achieve the expected.The results from the present studies suggest that the number recognition of electric meter can be completed by BP neural network, at one time, the recognition method of this studies is worth the design of automatic meter reading system learning.