Automatic Identification System Of Digital Meter Research And Design

With the development of computer technology and artificial intelligence, the management means and production modes of many industries gradually shifted from manual into automatic or semi-manual mode.During the shift, automatic identification system has played an important role, particularly in the harsh environments where data can't be directly accessed such as chemical industry, metallurgy and other industries. Automatic identification system can reduce the possibility of misjudgment caused by human factors, and improve productivity.Automatic identification system of digital meter is a non-contact real-time automatic identification system that based on machine vision , digital image processing technology and artificial intelligence. And it mainly including: video capture, image pre-processing, numbers region location, single digital segmentation, digital template library and digital recognition algorithm. The process is as follows: the digital meter monitoring through the USB camera or other video capture devices, image smoothing, gray-scale transformation, binarization, numbers region location and division of numbers into single character and so on. Aften all these preprocessings, the system identifies the data through the recognition algorithm, displays and saves the recognition results.Our research subject aim at the real-time data acquisition application needs of Henan Institute of Metrology for the PTW UNIDOS radiation dosimeter display. After a comprehensive comparison of multiple image processing methods and digital recognition algorithms and combination with the current image treatment and system performance requirements, we proposed new image pre-processing algorithms and digital identification algorithm,and develope a PTW UNIDOS radiation dosimeter automatic identification system. Main research contents are as follows:1, In the captured image, what really counts is the number of meter display. While the number regions is only a small part of whole image ,it is hard to achieve the desired effect using the global threshold. So we use the method based on local threshold processing to improve the quality of the image pre-processing.2, Through comparation and analysis of the regional positioning methods, considering radiation dosimeter PTW UNIDOS's own characteristics, and the speed, efficiency and accuracy, etc. , we select the manual-assisted positioning method locate a number of areas quickly and accurately.3, A new identification algorithm for system requirements is structured. Expanding the traditional template matching algorithm, we introduced a simple and easy way to implement template matching algorithm just to meet the requirement of high speed that a real-time processing system usually has.4, Because different digital meters have different foreground and background colors, the system should have a good adaptability.In our new system , different color components or gray-scale value ca be choosed as the basis for identification.