Indicator and sign recognition in real electricity scenario

Abstract A multitude of instruments and equipment are contained within the electricity scenario, and their operation leads to fluctuations in a wide range of parameters. These changes are typically represented through indicators and signs. This study introduces an efficient and robust method for recognizing these indicators and signs, utilizing a variety of image processing techniques, including the computation of image mean and variance, as well as template matching. Given the plethora of indicators and signs present on substation equipment, it is crucial to discern the illumination status of indicators at different locations and relay this information to the system based on the indicator’s significance. Considering the indoor setting, which is relatively uncomplicated, this study employs the variance and mean of the indicator area image for judgment. As for identifiers, which often vary in shape and color, this study uses template matching for identification, subsequently reporting the operational status of the equipment marked by this area. Experimental results demonstrate that compared to deep learning methods, traditional image processing algorithms yield superior outcomes while significantly reducing computational costs and time.