Building Compact Classifier for Large Character Set Recognition Using Discriminative Feature Extraction

In this paper, we propose an approach to building compact classifier for camera-based printed japanese recognition on mobile phones. We design feature vector prototypes using learning vector quantization(LV Q) for achieving high accuracy, while the complexity is lowered by the line ar dimensionality reduction. The descriminative feature extraction(DFE) strategy, which optimizes both subspace axes and classifier parameters, is shown to yield high classification accuracy even on low dimensional subspace. On a 120D subspace, a 4,344-class classifier consumes only 613KB storage, and an accuracy of 99.41% was obtained on test set.