Efficiency Analysis of Compared Normalization Methods for Fingerprint Image Enhancement

The efficiency of automated fingerprint identification system depends highly on fingerprint image enhancement algorithms. Fingerprint image enhancement can be divided into two stages: 1) At the first stage we enhance the image with conventional filters, where filter parameters do not change. 2) At the second stage we enhance the structure of ridges and valleys with contextual filters, where filter parameters change according to pre-calculated ridge orientation and frequency. This article presents an efficiency analysis of compared methods at the first stage with global normalization, local normalization and block local normalization, attempts to find the most efficient combination and evaluates the effectiveness of analysed methods from the perspective of their use in real time fingerprint identification systems. In the experimental part we compared three different types of normalization algorithms for enhancing image contrast, and then in combination with the second enhancement stage (Gabor filter and STFT) we assessed the enhancement system's effectiveness on database FVC2004.