S. Sae-Tang and I. Methaste (Thailand)
Thai Online Handwritten Character Recognition, Neural Network, and Windowing
This paper presents the development of Thai online handwritten character recognition system, which uses local and global features of Thai characters by applying the concept of feature feeding. A well-known multi-layer perceptron (MLP) neural network with backpropagation learning algorithm is chosen for its fast processing time and its good performance for pattern recognition problems. However, MLP requires that the network have a fixed amount of input nodes. Therefore, the windowing technique is developed to avoid the distortion caused by the normalization process. The input stream is sliced up by overlapping windows of the same size as the number of neural network input nodes. In applying with free-hand Thai single characters, the average recognition rate of 91.74% has been achieved this confirms that the proposed windowing approach is suitable for the development of Thai online handwritten character recognition system.
Important Links:
Go Back
