Analog measurement instruments, which do not have a digital interface, play nowadays an important role in industry. The calibration of analog measurement instruments not equipped with computer interfaces does not benefit from the advantages of automatic calibration. Therefore manual test methods are applied where the calibration technician must manually read the value of the reading of the instrument being calibrated, or must adjust manually the calibrator so that the instrument being calibrated presents some particular reading. It need to get a array data, which show the difference between practicality value and standard value at each grade marked. As we known, human vision system may lose accuracy after lots of vision work. These are very time consuming, and results are subject to human errors. This paper presents an automatic calibration system capable of analog calibrating measuring instruments. System tries to avoid the affects of checker's artificial mistakes such as vision errors and improves efficiency of inspection. Image analysis algorithms and image understanding algorithms are used to automatically determine the instrument reading. In order to obtain the reading of an analog instrument, it is necessary to determine the position of the pointer in the scale. The automatic determination of the instrument indication, from images of the display, requires the use of computer vision techniques.By using computer vision techniques, the pointer's positions in the scale of analog instruments are determined. A resolution much better than human visual acuity or the best analog instrument specifications was achieved. The use of such a system enhances the repeatability and accuracy of the calibration of measurement instruments that until now could not benefit from the advantages of automatic calibration.The application in this paper has several different modules: lighting, movement-organ, computer, program- control power, CCD capture system, machine vision software, database system and so on. The new movement-organ, which could perform a half-circle movement, has been used to ensure the system to trail the trace of pointer. Of course, the circumgyratetion centers are almost superposition point. When the pointer appears in the sight range of CCD, the use of machine vision technology enables the computer to read value through image which has been gotten by CCD image sensor, then drive the pointer to approach the grade which needs to be checked step by step by analyzing the relationship between pointer and the grade, the analyzing process was performed by the program- control power. When system satisfy some conditions, that is, the pointer had directed that grade exactly, the output value of program- control power is a inspect value which will save in computer and compare with a standard value. The difference between output value and standard value is a token of meter performance.There are some key technologies in this system: the design of system lighting, movement-organ to trace the pointer, arithmetic of machine vision to analyze image. As this system work in constant temperature room, some light source in room distribute random which cause disequilibrium of luminance on meter surface .In that system, as for the cause mentioned before, a lighting device simulate light conditions outdoor to avoid affect of noise light .At the base of light conditions computer control the movement-organ which could drive CCD image sensor to move along the trace of needle, then capture image needed at compatible time .The arithmetic of machine vision is most important one. This system pre-process image to gain a 768*568 gray image firstly, then use dynamic thresholding technology, which reduce background noise and white noise, to enhance information of needle and grades. And then, by make use of connected component technology, system get needle image with some grade, where computer read value through analyzing the relationship between needle and the grade. Some experimentations show that this system have a practicality... |