Automatic Pointer Meter Reading Based on Machine Vision

Pointer meters in power systems are wildly used. Although smart meters gradually replace pointer meters to conveniently acquire measurement, digitalization of some special meters have not been available. The paper presents a new method based on machine vision to detect pointer meter reading automatically. Two phases of training and testing are proposed in this method. In the training phase, a template of meter feature points on a training image is extracted. In the testing phase, the template matching performs on a testing image to find feature points. Then, an elliptic arc fitting based on the feature points to generate a 1-D signal. Finally, the 1-D signal is used to locate the pointer and the reading result is obtained. Differing with existing methods, the proposed method performs effectively and efficiently automatic meter reading without using Hough transformation.