Handwritten character recognition using steerable filters and neural networks
S. Talleux, V. Tavşanoglu, Emir Tufan · 1998
In this paper a system for handwritten character recognition has been developed using a steerable filter and a neural network where the former yields the local direction of dominant orientation while the latter, using this feature, recognises the character. Two different models of neural networks have been tested: (i) the single layer perceptron with error correction and (ii) multilayer perceptron with backpropagation algorithm. The pre-processing consists of several stages.