Electric Power Meter Classification Based on BOW

Abstract

The automatic verification of power meters is of great significance, and the key point is the classification of the power meter types. In this paper, we propose a power meter type recognition method based on machine vision. We construct a Bag-of-Words model(BOW), and extract the image features of the instrument, and construct a visual dictionary, based on which to train a support vector machine classifier to realize the automatic identification of the instrument type. The experimental results show that the proposed method achieves a classifiaction rate of 100% for several specific power meters, and is of great significance for applications.