Meter Image Recognition Technology Single Phase

In the electric industry, the management of electric consumption and the information input of damaged electric meters are traditionally conducted by workers. To overcome traditional method’s high complexity and low efficiency, an intellectual electric consumption recording system is required. An automatic recognition system which consists of images collection, processing and recognition for single-phase electric meters’s parameters was discussed and developed in this thesis. This research proposes electric power utilities an efficient tool when acquiring users’ electric consumption data and preserving electric meters’ status information. Below are this thesis’ main contents:(1)Image preprocessing technique of electric meter was studied. The median filtering technique was used to smooth colored images. With the Sobel image edge detection operation, images’G parameter was abstracted for level edge detection. Hough transform method was adopted to rectify the image’s inclination with bilinear interpolation. With a consideration of LCD and barcode area’s characteristics, the Hough transform method was used to locate LCD output area and the sweep-line method was applied to find out the bar-code’s location.(2)The Bernsen algorithm was optimized to separate well the target and background under non-uniform illumination. With an analysis of character type LCD’s features, the recognition rules were defined for electric meter’s LCD characters. The characters were successfully recognized together with the structure characteristic algorithm.(3)According to the different size of black and white strips, the bar code’s binarization was realized with the OTSU global threshold method and the improved Bernsen algorithm. After a study of Codel28’s coding rules, the accuracy of bar code’s recognition algorithm was improved by extracting and rectifying the effective lines. The code words were finally read by calculating the distance of edge to similar edge.(4)A system for the single-phase electric meter based PC was realized. The operation process of this system was designed. An integrated software was developed for image real-time acquisition and recognition. This software’s performance was comfirmed by test.