Image Processing Technique for Improving the Sensitivity of Mechanical Register Water Meters to Very Small Leaks.

Discovering very small water leaks at the household level is one of the most challenging goals of smart metering. While many solutions for sudden leakage detection have been proposed to date, the small leaks are still giving researchers a hard time. Even if some devices can be found on the market, their capability to detect a water leakage barely reaches the sensitivity of the employed mechanical water meter, which was not designed for detecting small water leakages. This paper proposes a technique for improving the sensitivity of the mechanical register water meters. By implementing this technique in a suitable electronic add-on device, the improved sensitivity could detect very small leaks. This add-on device continuously acquires the mechanical register's digital images and, thanks to suitable image processing techniques and metrics, allows very small flows to be detected even if lower than the meter starting flow rate. Experimental tests were performed on two types of mechanical water meters, multijet and piston, whose starting flow rates are 8 L/h and 1 L/h, respectively. Results were very interesting in the leakage range of [1.0, 10.0] L/h for the multijet and even in the range [0.25, 1.00] L/h for the piston meter.