Palmprint Recognition Research Based On Improved PCA And ICA

Biometrics is a technique,that through the computer with a variety of sensors,biometric principle and other high-tech technique closely integrated, use the inherent human physiological characteristics and behavioral characteristics, to carry out the identification of personal status. Now it is considered as the highest-ranking security key system,in the height of the internet information-based society.Palmprint Identification refers to the human effective information of the palmprint to identify individuals. The human palmprint have be exclusive for others, and invariability, according to the feature, palmprints of the human body can be used as identification.There are many algorithms in Palmprint identification, including applying point characteristics of palmprint, frequency-domain methods, principal component analysis, the independent component analysis, and so on. All of which have their own advantages and disadvantages. Thus in recent years, research scholars tend to be mixed several algorithms together to achieve palmprint identification.This paper is starting for this point,research PCA pivotal technology of the palmprint identification technology with great concentration and painstaking, and take improvement and innovation on the basis of their algorithm. That is aiming at that classical PCA algorithm is not applicable to the classification, and the experiments in this paper, classifying those eigenvalue that take sample dispersion within the category of smaller, and take sample dispersion between the category of larger,to describe that the original palmprint images.Following experiments show that this improved algorithm will not be lost original information, and to a certain extent enhance the recognition rate. The final experiment combine the improved PCA algorithm with ICA algorithm,that is after disposaling of improved PCA algorithm,take the extraction of palmprint images in Independent Component Component.Experiment data show that this method not only has advantages in terms of time,but complete identification with less time for the same palmprint images, and the recognition rate was significantly enhanced.