Hand Vein Recognition Based On Reference Point And NMI

With the continuous progress of modern society,science and technology,how to protect the information more effectively is especially important.Since traditional recognition methods widely used in daily life(such as,password input,intelligent IC card and even identity card,etc.)have deficiencies in the aspect of security and convenience,the technology that could recognize individual real identity effectively is urgently needed.Using body innate physiological characteristics to recognize identities,biological recognition technology has been developed a lot in recent years.Compared with other biological characteristics,such as fingerprints,DNA,irises,facial images and so on,hand vein recognition technology stands out depending on many advantages,like stable characteristics,uncopy,non-contact,simple sampling device,etc.In this paper,the processes that vein images' collection,preprocessing,feature extracting and recognition are discussed in detail.Besides,a hand vein recognition algorithm based on reference point and NMI is proposed in this paper,which can enhance recognition's accuracy rate.The important procedures of this paper are shown as follows:1)Veins images' collection system is established,and the core equipment of this system(light source,imaging device,etc.)is analyzed.The system we build has practical values with simple structure,strong handleability,high-speed collection and high receptivity to users.2)Before recognition,vein images are preprocessed.The algorithms about positions' correction,denoising,enhancement,segmentation and refining are studied,and the preprocess results of vein images are presented.3)A hand vein recognition algorithm based on reference points and NMI is proposed in this paper.First,reference points of vein images are determined.Then,the number of the same included angles between the two adjacent lines of reference point and feature points is counted and compared.And then general NMI values and sub-block NMI values of vein images are compared separately,so the identification of vein images could be completed rapidly.This method has not only strong antitypy to the effects caused by image translation and rotation,but also good instantaneity,which improves recognition accuracy rate a lot.Comparison of the algorithm proposed and the other methods(area code and invariant moment),experiment results testifies validity of this paper's algorithm.