Low-quality fingerprint recognition using three- rate hybrid Kohonen neural network

One of the most difficult problems in fingerprint recognition has been that the recognition performance is significantly influenced by fingertip surface difficult conditions (e.g., rough fingertips, dry fingertips, allergic-skin fingertips), which may vary depending on environmental or personal causes. Addressing this problem, this paper presents the three-rate hybrid Kohonen neural network (TRHKNN) for low-quality fingerprint image processing. The received TRHKNN consists of "fast" Kohonen neural network, "middle" Kohonen neural network and "slow" Kohonen neural network. The illustrative example - recognition and restoration of distorted low-quality fingerprint - was carried out using this TRHKNN. The received TRHKNN has not only high speed of image recognition, but also high speed of image restoration. The simulations for the TRHKNN were carried out in a MATLAB/Simulink environment. This example demonstrates an efficient recognition performance of the proposed computing procedure and applicability of the TRHKNN for fast-acting low-quality image recognition in real-time conditions.