Hybrid fuzzy feature extraction framework for handwritten numeric fields recognition

A hybrid feature extraction framework for handwritten numeric fields recognition is described. The numeric fields were extracted from binary images of credit card application forms. The images include identity numbers (ID) and telephone numbers. The feature extraction framework utilizes a cascade of multiple Kohonen self-organizing feature maps(SOMs) and sets of membership value generation units. The goal of our feature extraction process is to provide reliable information to the recognition stage. The recognition stage uses the fuzzy feature set as inputs to a multi-layer neural network. The desired outputs for the networks were set using a fuzzy k-nearest neighbor algorithm. We present experimental results which demonstrate the ability to extract features automatically in handwritten digit recognition. Experiments were performed on a test data set from the CCL/ITRI Database which consists of over 90,390 handwritten numeric digits. A recognition rate of 98.74% is achieved on this database.