Efficient Way of Reading Rotary Dial Utility Meter Using Image Processing

Reading a rotary dial utility meter can be difficult and, more importantly, inaccurate. This paper has developed a computationally efficient method to read a rotary dial utility meter via image processing. The method has been developed as an algorithm with the ultimate goal of implementing the system with a microprocessor and inexpensive camera. The algorithm registers successive dial images, subtracts them to detect dial movement, and uses pattern recognition to locate each dial. The movement of each dial is then used to calculate utility consumption. Sample meter images were obtained by scanning an actual rotary meter. The algorithm's image resolution requirements were determined. The algorithm was simulated in Matlab and timing analysis performed to determine where the algorithm may be improved. The algorithm was successfully run with sample meter images with the algorithm correctly determining the indicated consumption.