NEURAL NETWORKS AND HAUSDORFF DISTANCE APPLIED TO NUMBER RECOGNITION IN ELECTRICAL METERS

摘要

Pattern recognition is the area of research dedicated to recognizing objects, images, faces, letters, numbers, and so forth. Number recognition processes have an important role in remotely monitoring data for electrical meter readings, and monitoring data from these devices can help to reduce energy consumption. Research in the area of number recognition is vast and there are many different methods have been developed; some of these approaches follow characteristics extraction methods and others, such as the Hausdorff Distance, use the calculation of the distance between two finite sets. In this article, some of these approaches and a comparison among them are presented. Results showed that for recognizing complete digits, characteristic extraction methods offer a better result in terms of recognition time than Hausdorff Distance methods; however, both are similar when considering recognition percentage.