Offline signature recognition system using histogram of oriented gradients

The most common way of authenticating a document or financial transactions, especially cheques, is handwritten signatures. In most of the cases, verification of these signatures is done by visual inspection. A person compares the two signatures and accepts the given signature if it sufficiently matches with stored signature. In banks where thousands of cheques and scanned documents are to be processed every day, process of visually verifying the signatures become cumbersome and time consuming. Automating the signature verification will improve the situation and eliminate the possibility of forging. In this paper, an offline signature recognition system which uses histogram of oriented gradients is presented. Feedforward backpropagation neural network is used for classification. The system gives recognition rate of96.87% with 4 training sample per individual.