Automatic Power Consumption Reading on Digital Meters based on Adaptive Thresholds and Multiresolution Templates

The reading of electricity consumption in Brazil is usually performed manually. One way to improve the time spent in the reading process is to provide a tool that recognizes the consumption based on images. In this context, this paper presents a method based on adaptive template matching for recognizing the power consumption displayed in digital power meters trough images. This is a part of SILEM - Mobile Reading System module, a validation system to automatically read and process images that are captured in the field, detect inconsistencies and at the same time perform automatic consumption recognition. The methodology results are promising, it recognized 72.18% of all meter consumption correctly, with 86.38% of all meter having at least 5 of 6 digits correctly recognized, minimizing human interactions to correct the read power consumption.