Comparison of Two Algorithms for Automated Multi-Meter Testing using Character Recognition
Character and digit recognition is a basic building block in many automation tasks now-a-days. For keeping records in computer and digital systems, data in digital format is required. Characters and digits recognition is done with the help of artificial neural network. This paper proposes comparison of back-propagation algorithm and K-Nearest Neighbor algorithm that recognizes digits and characters on digital display. This includes different techniques i.e. preprocessing, feature extraction, training and testing of data in the system. Comparison of algorithms can be done on different criteria’s such as accuracy, power requirement, time required for training and testing, system complexity for different set of image dataset.
Keywords - Artificial Neural Network, Back-Propagation Preprocessing, Epochs, K-Nearest Neighbors, Training Error Rate