Summary: | 碩士 === 國立高雄應用科技大學 === 機械與精密工程研究所 === 96 === The traditional meter manufacturing plants adopt manual inspection mostly. However, fatigue of manual operation and time-consuming lead to poor quality of meter calibration. By using machine vision method, inspection quality can be increased and inspection time can be reduced. This study focused mainly on home watt meter calibration, such as dial indictor, watt meter number, numerical indictor and rotor indicator, by using automatic detection, acquisition, identification and inspection.
Dial indictor detection is based on the pixel gray level average and standard deviation. Watt meter numbers and numerical indictor recognition is based on Model Finder algorithm of Matrox Image Library. By calculating the pixel number of binary image and judgment rule, the rotor indicator is counted.
Experimental tests of the proposed system show successful rate of 100% for dial indicator, 80% for numerical indictor, and 90% for watt meter number. In the verification of dynamic rotor, this system also provided a better solution approach.
|