HMM-Based Off-Line Handwritten Chinese Characters Recognition Using Krawtchouk Moments

This paper proposes a new efficient approach using the set of Krawtchouk moments and HMM for character recognition. The Krawtchouk moments set is a new set of orthogonal moments based on the discrete classical Krawtchouk polynomials. Compared to continuous moments such as geometric, Hu invariants, Zernike and Legendre moments, the implementation of Krawtchouk moments doesn't involve any numerical approximation and coordinates transform, since the basis set is orthogonal in the discrete domain of the image coordinates space. This property makes Krawtchouk moments superior to the conventional continuous moments. In character recognition field, HMM is a good solution because of its strong adaptive to the written deforming. In this paper, Krawtchouk moments and HMM are combined to recognize off-line handwritten legal amount. Experiments confirm the performance of our proposed methods. The average recognition is much higher than Legendre moments and Hu invariants under HMM framework.