Dorsal hand vein biometric using Independent Component Analysis (ICA)

Lately, dorsal hand vein pattern is gaining popularity in biometric security system due to its uniqueness and stability. Though dorsal vein patterns are not complex, this does not reflect in its extraction and representation. Various existing methods consider vein pattern as straight lines or use some of its features like ending points and bifurcation points for its representation. However, this type of vein representation is not the ideal solution since important features may be lost. Consequently, using all the pixel values representing the vein pattern is a better alternative. But the processing and matching time is at stake when using all the pixel values of the considered image. This problem can be solved by using dimension reduction techniques. In this work, the vein patterns are represented using Independent Component Analysis (ICA). This representation and reduction technique is carefully explored by providing details of the vein matrices. The experiments are carried out on 100 individuals, and the efficiency of the system is tested by the computation of the false acceptance rate and the false rejection rate.