Skip to Main Content
This paper presents a novel approach to recognize similar handwritten numerals based on empirical mode decomposition (EMD). We firstly use the local maximum modulus of wavelet transform (MMWT) to get the width-invariant and grey-level invariant characterization of contours in an image. Then we apply EMD analysis to decompose the synthetic shift normalization of curvature into their components, which could produce more compact features. Finally, three different classifiers, i.e. support vector machine (SVM), hidden Markov model (HMM), and artificial neural network (ANN), are used to discriminate similar handwritten numerals for testing the effectiveness of the extracted features. Experimental results show that the proposed approach obtains higher recognition rates compared with the traditional algorithm for extracting features.