Application of Dimensionality Reduction Analysis to Fingerprint Recognition

Dimensionality reduction is an important issue in Fingerprint recognition that often faces high-dimensional data. Two-dimensional principal component analysis (2DPCA) is one of the most popular methods for dimensionality reduction. A novel fingerprint recognition algorithm using 2DPCA has been proposed in this paper. Firstly, the prime features of original images can be attained by two-lever WT decomposition. Secondly, the features of dimensional reduction are solved by 2DPCA. Finally, fingerprint recognition can be realized by Ellipsoidal Basis Function Neural Network (EBFNN). The algorithm combines the optimization of the 2DPCA and the adaptability of EBFNN. The resulting algorithm is tested on three different fingerprint verification challenge datasets and demonstrates much higher performance in comparison to WT-2DPCA-RBF.