基于图像识别的电能表智能建档系统研究

目前,省级计量检定中心正在向智能化、无人化、高可靠方向发展,但在电能计量设备到货建档环节仍采用人工维护方式。鉴于此,依据电能表主要建档参数均印刻在电能表表面、位置相对固定的特点,开展了基于图像识别的电能表智能建档技术研究,提出了基于SIFT特征区匹配方法、基于大津法的改进垂直投影直方图方法以及基于Bi-gram语言模型的电能表参数串识别后处理方法。应用上述技术研制了电能表智能建档系统,实现了单相电能表参数、三相电能表参数的自动识别和建档等功能。通过实验,验证了电能表智能建档系统可准确定位、切分和识别建档参数,为计量设备智能建档的推广应用提供理论和实践指导。

Research on Intelligent Archiving System of Energy Meter Based on Image Recognition

At present, the provincial metrological verification center is developing towards intelligent, unmanned, and highly relia-ble. However, the manual maintenance method is still adopted in the electrical energy metering equipment arrival ar-chiving. Therefore, based on the characteristics of the main parameters of the energy meter file being engraved on the surface of the meter and the relatively fixed position, the research of intelligent file archiving technology based on im-age recognition was carried out. An improved vertical projection based on the feature region matching method and the Otsu method was proposed. Histogram method and language model-based post-identification method of electric ener-gy meter parameter string. The above-mentioned technology is used to develop an intelligent energy meter archiving system, which realizes the functions of automatic identification and archiving of single-phase meter and three-phase meter parameters. Through experiments, it is verified that the smart metering system can accurately locate, segment and identify the parameters of the filing, and provide theoretical and practical guidance for the popularization and application of metering equipment intelligent filing.