The Research Of Wheel-Power-Meter Reading System Based On Digital Image Recognition

With rapid development of the efficiency of market regulation, automatic meter reading (AMR) system has become one of these indispensible links, which could realize the automation of measuring and supervising electric power. But the fact that updating to current AMR systems needs to refit or replace old power meters is not suitable for old users of rual ereas, leading to still prevalence of manual meter reading. To solve this problem, a wheel-power-meter reading system on the basis of digital image recognition, accomplishing automatic acquisition and process with features like modification-free, replacement-free, and contact-free, was put forward in the paper.In this paper, the features of digits on the wheel-power-meter and techonolgies of digital image recognition were deeply studied for the design of wheel-power-meter reading system on the basis of digital image recognition. In the system, the acquisition of image from the wheel-power-meter, automatic recognition of power value were accomplished, and so were management functions, such as saving, charging, query, revising and so on, for relevant information like power, users and so forth.In this paper, the digital recognition arithmetic was studied and designed based on digital image pre-processing, digital feature extraction, and digital feature recognition. In the study of digital image pre-processing, the arithmetic for accurate locating digits' regional images on the basis of location of the wheel region image was put forward in this paper, to obtain the accurate location of all digits; and so was the arithmetic for information strengthening in digital images to extract the skeleton information of digits. What were emphasisly studied are locating arithmetic based on the extraction of edge contour and its rectangle feature, different binary conversion arithmetic suitable for locations of all levels and information strengthening, and revised arithmetic under the incomplete situation of the sixth digit. Then, arithmetic for extracting structural features under nine grids was studied and designed from aspects of closed hole, branch point, and stroke; and so was feature recognizer with the digit having maximum accumulated weights as its ouput. The software of this system was built by C# language, which was tested and verified using image samples.This system accomplished image collection, digital recognition, and intelligent management of the numerous information of the wheel-power-meter, which was demonstrated precise and reliable by experimental results.