Table 11 Comparisons with other methods without feature selection
From: A hybrid recognition system for off-line handwritten characters
Authors | Dataset | Feature extraction method | Classifier | % Accuracy capital alphabet | % Accuracy small alphabet |
|---|---|---|---|---|---|
Murthy and Hanmandlu (2011) | Samples From Matlab | Directional features | Fuzzy logic | 83.85 | Â |
Nasien et al. (2010) | NIST | Freeman chain code | Support vector machine | 88.46 | 86.00 |
Ali et al. (2010) | – | Wavelet compression | Euclidean distance | 89.68 |  |
Hanmandlu et al. (1999) | – | Fusion method | Neural network | 86.00 |  |
Singh and Hewitt (2000) | CEDAR | Hough transform | LDA Nearest neighbour | 67.3 63.5 | Â |
Vamvakas et al. (2009) | CEDAR | Structural features | SVM | 80.19 | Â |
Ganpathy and Liew (2008) | Self Created | – | Multiscale neural network | 87.22 |  |
Yuan et al. (2012) | UNIPEN | Â | Convolution Neural network | 93.7 | 90.2 |
Our approach | CEDAR | Hybrid features | Neural network | 91.56 | 87.49 |