Palmprint Identification Using DCT-based Multi-color Component Images Fusion

With the rapid development in the information, digital technologies, and network of modern society, the traditional methods like password can't meet the needs of security authentication. The corresponding biometrics identification technologies have been the trend on human identification. Palmprint recognition which is used widely on information security and criminal evidence investigation receives much concern because of its advantages.One important research contents of palmprint recognition is finding distinguished features. Using color information in palmprint recognition is described in this paper, including single color component image and multi-component images after analyzing the existing papers about extracting palmprint feature methods. A good experiment result is obtained on low resolution color palmprint images. The main works of the paper are as follows:(1) An improved DCT based palmprint recognition algorithm is proposed in the paper. The algorithm gets a better result in feature vector dimension reduced, recognition rate enhanced and recognition time shortened comparing with PCA/wo3 algorithm and 2D-DCT algorithm.(2) The paper uses the color information in palmprint recognition after summarizing and analyzing the main existing papers on extracting palmprint features. The experiment is carried on color palmprint images. The recognition results on color component images, such as R component of RGB color space, I component of YIQ color space and S component of HSI color space.(3) Using multi-component images feature level fusion in palmprint recognition is proposed in the paper. And this method gets a better experiment results on recognition rate and EER comparing with using single color component image.