Dimension reduction of the feature vector of Liquid Drop Fingerprint

In order to decrease the cross-validation time and improve computation efficiency, this paper presents a method to reduce the feature vector dimension of Liquid Drop Fingerprint (LDF) by using factor analysis and principal component extraction. Waveform analysis is one of the best in many feature extraction methods. It can grasp the main features of LDF, but the feature vector reaches up to 10 dimensions. And the problem of information overlap in feature vector adds unnecessary computational complexity for pattern recognition. The calculation shows that the feature vector dimension is reduced to 30% by factor analysis or principal component extraction, while the recognition rate decreased less than 4% and the cross-validation time is reduced by more than 64%, the computing efficiency is improved remarkably.