Research On The Reading Recognition Method Of Smart Meter Based On Image Processing

Nowadays,electric energy is an indispensable energy source in daily life and industrial production;smart meters,as a new instrument for measuring electric energy,have been widely popularized.Due to the high cost of the traditional manual meter reading method and the limitations in extreme environments such as high temperature and high pressure,scholars have begun to study the use of image processing technology to achieve remote reading of electric meters,which has realized digital extraction,segmentation,identification,and It can get a better recognition rate under various constraints.However,due to the variety of instrument styles and complex use environment,it is difficult to ensure that the identification system maintains high-precision readings during transplantation.Based on the above reasons,this article deeply studies the existing domestic and foreign LCD screen instrument recognition algorithms,and studies the limitations and shortcomings of the existing algorithms,Taking smart meters as the research object,this paper proposes a meter reading recognition method based on digital image processing,which aims to identify the meter readings quickly and accurately,and can obtain a higher recognition rate under the influence of harsher environments.The main research contents are:1.Preprocess the collected images.Because the size of the images collected by the camera is different,the images are normalized and adjusted to a uniform size;In order to reduce the amount of calculation and save storage space,it is necessary to grayscale the color image.At the same time,due to factors such as image acquisition,transmission and environmental influences,the image will be interfered.Bilateral filtering is used to eliminate these interferences,a grayscale image with uniform size and clear image is obtained.2.Positioning the LCD(Liquid Crystal Display)screen area,eliminate interference outside the LCD screen.In order to realize the LCD screen positioning,this paper analyzes the problems based on the edge positioning algorithm,and designs an algorithm that combines the edge positioning and the Hough linear transformation to locate the screen twice and perform tilt correction.First convert the image to the HSV domain,and select the S-channel image for binarization.After the binarized image is subjected to morphological operations,the image outline is extracted,and the image is filtered according to the LCD screen outline characteristics and the outline is intercepted to complete the initial positioning;Then use the improved Canny edge detection and Hough line detection to complete the tilt correction and precise positioning of the screen.Experiments have verified that the screen segmentation accuracy of this algorithm reaches 98%.3.Threshold the image to realize the separation of the target and the background and the recognition of digital characters.By analyzing the existing binarization algorithm and combining the global threshold method with the local threshold method,a threshold segmentation method combining OTSU algorithm and Sauvola algorithm is designed,this algorithm can effectively solve the problem of uneven illumination and darker light in the image,while greatly improving the processing speed.In the digital recognition part,the projection method is used to locate and segment the display content of the LCD screen,and the digital characters are screened by morphological characteristics,and finally the number recognition is completed by the threading method.In the last experimental part of this article,300 pictures in different situations are used to test this method and Sauvola algorithm.The final recognition accuracy rate is97.33%,and the average processing speed is 1.408 s per image and are better than Sauvola algorithm with 86% accuracy rate and 2.770 s processing speed per image It has great application value in daily meter reading and industrial production.