Research And Implementation On Face Recognition System

Nowadays, with the rapid development of computer technology and science,people even pay more attentions aboat the information in the field of security. But the inaccuracy and irreplaceable of the traditional identification methods obviously can not greatly meet the demand of contemporary society. Therefore, the technology based on their biometric to identify the identity is emerging. Such as palm fingerprint, iris and retina of the eye as well as the human diet(handwriting, shape) etc. Face recognition technology which is different from other biometric technologies has the characteristics aboat easy identification and the only characteristic determining. You can complete the identification work separately from the active-passive in two ways.So it gets the in-depth research of scholars and a wide range of applications on markets,and self-evident has become the field of information flurry of biometric technologyThrough research and analysis comparison of the existing face recognition technology, combined with their advantages and disadvantages, I determine using two-dimensional principal component analysis(2DPCA) and Gabor wavelet transform of these two algorithms,and make the in-depth research on extracting its features.Then we make a detail on the main idea and their characteristics as well as the processes and the achievement on feature extraction aboat the two algorithms. Then this paper make a simple improvements on face detection algorithm Adaboost algorithm, it improves the accuracy and detection accuracy, and have been made improvements on feature extraction and classifier design.Firstly, let’s build a global classifier by 2DPCA global feature extraction, then using two-dimensional Gabor wavelet transform extracte the local features and building multiple local classifier based on spatial location. Finally, it constitutes the final overall classification as the weighted sum of the way parallel integration,and conducts simulation experiments on International General ORL database and self Face library. Experimental results show that, this paper proposes two methods of face recognition algorithm combining can reach 97% on the premise of no influence on the speed of face recognition,and it achieves the desired goal.