Artificial database expansion based on hand position variability for palm vein biometric system

Biometrics is of utmost importance in security systems that are also convenient in use. The ones based on palm vein approach are not widely known yet but they seem to overcome disadvantages of popular methods. In this paper a palm vein system is proposed. In addition, the method of elastic deformation which is used in order to create a membership function that models hand position variability is shown. This approach allowed the improvement of the results by 3.78 % on the average in the comparison to the outcome of the matching done using Local Binary Patterns feature vectors without the membership function. The True Positive Rate without the membership function was already on the level of 92.42 %. This proves that the built system performed well as it comes to the palm vein recognition and that the proposed way of obtaining the membership function leads to the improvement of overall classification results.