Handwritten alphabets recognition system using Radial Basis Function Network / Mohd Afandi Yob

This thesis presents how to recognize handwritten alphabets using Radial Basis Function (RBF) Network. This Radial Basis Function is an information processing paradigm that is inspired by the way biological nervous systems, such as the brain and process information. RBF Networks has the learning process that contains training and testing processes to obtain the outputs. The learning process involves additional delta weights that include RBF Networks parameters. This process is important to find the optimum weights of the RBF Networks. Handwritten alphabets were taken from 10 sets of samples that contain 260 different types of characters. These handwritten alphabets were used to be as inputs of Radial Basis Functions (RBF) for training and testing process. Using MATLAB programming, these data was trained for the RBF Networks through training process that include finding the center of RBF, hidden layer, delta weights and finally to get optimum weights. From optimum weights, the output was obtained. During testing process, other handwritten alphabets were tested. Three samples of handwritten alphabets are tested using Graphical User Interface (GUI).Using the same center as training process, RBF Networks calculate the input data and recognize them as an alphabets.Using programming in MATLAB software, the outputs of Neural Network system was successfully obtained programmatically. This system is important for recognizing different types of handwritten alphabets. As conclusion, this system handwritten alphabet can be recognized successfully