Research On Face Recognition Technology Based On 2DGabor Wavelet Transform And 2DPCA

Gabor wavelet transform is a kind of good face description method. The characteristic of 2DGabor is more robust than of the simple gray feature. PCA method is an important means to reduce the dimension of face image. 2DPCA method avoided conversion from the image matrix to the one-dimensional vector and obtained satisfactory results in face recognition. For robustness and practical application in face recognition technology, a combined face recognition method is proposed to base on 2DGabor wavelet transform and 2DPCA.Firstly, Gabor wavelet transform and PCA technology were outlined. And these technologies for face recognition application were given a comprehensive presentation in this article.Secondly, a new face recognition method was proposed to combine 2DGabor wavelet transform with 2DPCA by analysis and study of the two technologies. As the problem that features matrix exist high-dimensional in face recognition technology, the feature integrated-method of the PCA was raised to relate the characteristics of a two-way and two-dimensional PCA with two-dimensional PCA, thus this method will significantly reduce the dimension of features matrix. In order to further improve speed of recognition and accuracy of recognition, integrated-approach is proposed to base on dynamic weight of BP neural network, which effectively improve the speed and accuracy of identification for face recognition system.Finally, face recognition system was designed and realized by the combined-algorithm which combined 2DGabor with 2DPCA. Experimental results in MATLAB show that the combined-algorithm is better than algorithm that combined Gabor wavelet transform with DPCA for recognition effect.