A Neural Network-Based Handwriting Recognition System

• Main Authors: Dinh Hai Bang, 丁海朋
• Other Authors: 徐元寶
• Format: Others
• Language: en_US
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/g6524w

碩士 === 國立虎尾科技大學 === 資訊工程系碩士班 === 104 === In this thesis, we design and implement a handwriting character recognition system for Vietnamese alphabets. A neural network-based algorithm of handwriting recognition is derived. Simulation is conducted by applying the algorithm to a virtual machine to verify that the method is able to perform the recognition. The system is realized on an Android smart phone. The handwriting images are extracted from the mobile device’s touch panel, then these images are processed through segmentation and wavelet transform feature extraction techniques. After that, a feed-forward neural network is applied for character recognition, which improves the accuracy significantly. The resulting system is operated smoothly with 90% of accuracy.