Sparse Representation Based on K-Nearest Neighbor Classifier for Degraded Chinese Character Recognition

摘要

In this paper, we present an effective coarse-to-fine algorithm to recognize the degraded Chinese characters. The algorithm contains two basic steps. Firstly, for the test images and the train images, reduce the dimension of the character feature via principal component analysis (PCA), and K-nearest neighbor classifier is exploited to find the candidate recognition results. Secondly, a sparse representation algorithm is explored as a fine recognition classifier. A dictionary is constructed by the PCA feature spaces of all the training images of the candidates' categories to reconstruct the input image via sparse representation, and the residual error is calculated by the sparse coefficients corresponding to each candidate category. We apply the method to the low resolution and noised 3755 categories of Chinese characters, the comparison experiments verify the efficacy of the proposed algorithm.