For the problem of high complexity and misrecognition rate in face recognition algorithm,a new method by fusion of SVM and AdaBoost is proposed based on 2-dimensional principal component analysis（2DPCA）.Firstly,the algorithm produces “feature face” through face detection,wavelet transform and 2DPCA to near-infrared image.Then,feature data is learned by SVM to build initial classifiers.They will become strong classifiers when reinforced by AdaBoost.Finally,strong classifiers act on test samples to complete the identification.Experiments show that the algorithm not only increases the ability of classifier,but also reduces the computation complexity.Meanwhile it improves recognition probability in practical application.