This paper proposed a novel Handwriting Thai Signature Recognition algorithm, we used two groups of features. Global feature such as image area, pure height, pure width, center of handwriting Thai signature in the vertical, center of signature, horizontal, etc. and Grid feature with three block size then take all the features into a neural network for training for signature images from 10 persons with each 60 images by calculating the average of the signature of each person from all database for represent.
When through the process of learning with Multilayer perceptron (MLP) trained using back propagation algorithm and then fed to the neural network is another layer Radial basis Function (RBF) for use in decision.