Abstract:To overcome the instability of handwritten character caused by different writing style, a novel approach of model reconstruction based on manifold learning is proposed in this paper. Traditional recognition method reduce dimension firstly, and then recognize the character in the reduced feature space. In this paper, we present a method based on reconstruction LLE. The algorithm first reduce dimension in each class. Then reconstruct the character in each class. Finally character recognition is then conducted based on the error analysis of reconstruction model. The algorithm proposed in this paper is tested on the characters in MINST character database and the experimental results demonstrate that the method can effectively improve the recognition rate of handwritten digits and provide a new approach to the research of handwritten digits recognition.

Handwritten Numeral Recognition by Model Reconstruction Based on Manifold Learning