Adaptive Normalization of Handwritten Characters Using Global/Local Affine Transformation
IEEE Transactions on Pattern Analysis and Machine Intelligence
Hi-index | 0.00 |
For offline handwritten Chinese character recognition, stroke variation is the most difficult problem to be solved. A new method of optimal sampling features is proposed to compensate for the stroke variations and decrease the within-class pattern variability. In this method, we propose the concept of sampling features based on directional features that axe widely used in offline Chinese character recognition. Optimal sampling features are then developed from sampling features by displacing the sampling positions under an optimal criterion. The algorithm for extracting optimal sampling features is proposed. The effectiveness of this method is widely tested using the Tsinghua University database (THCHR).