An Energy Visualization by Camera Monitoring

This paper proposes an energy visualization system by a camera. For monitoring, a single camera captures gas meter image at fixed intervals. The system applies edge detection and the connected-component labeling to extract numeral regions in counters of a gas mater. Gas consumption is estimated based on shape characteristics of numerals. The system uses number of endpoints and holes in numerical character, and it calculates a direction histogram and sum of absolute difference (SAD). The system recognizes the numeral by fuzzy inference from the acquired shape characteristic. When the system failed to recognize gas consumption by some accidents, the consumption is interpolated from time-serious data. In the result, our method estimated 32 and 29 numerals in 33 pieces for front and slant measurement respectively. For a continual monitoring in a day, the system successfully estimated dynamic gas consumption change and visualized them.