Researches Of Face Recognition And Image Alignment Based On Sparse Representation And Low Rank Matrix Decomposition

In recent years compressive sensing theory and algorithm have been proposed. Sparse representation and low rank matrix decomposition success in high similarity image classification and recognition. In this paper, we combined the two methods then proposed a new algorithm of face alignment and recognition base on sparse representation and low rank matrix decomposition. According this method, we promote the algorithm of low rank matrix decomposition in face recognition. In addition, due to the theory of low rank matrix decomposition can be used for image alignment. Try to apply this method into skew scanned document images in order to study whether this method can align the skew scanned document images correctly.Due to Sparse Representation Classification basing face recognition algorithm is easily disturbed by occlusion, noise and misaligned. It is difficult to obtain a perfect performance, for this, a new method is proposed in this paper. First the training matrix is decomposed into a low rank matrix which is the clean face image and a sparse error matrix representing the noise, occlusion and other errors. Then a transformation matrix factor is utilized in the optimization model, which can be computed while decomposing the training matrix, realizing the auto face alignment in XY plane. Last, the low rank matrix is used as the face training data to be classified by sparse representation method. Experimental results show that our method can align the face images and perform better than traditional face recognition method.In the aspect of application, due to low rank matrix decomposition can align the low rank images automatic. In reality, the scanned document images are easy to shift. But the rank of the scanned document images may be low. So we can use low rank matrix decomposition to align the images. Decompose the original matrix into low rank matrix and sparse error matrix, and in the process of decomposition a transformation factor would be obtained, use the transformation factor to align the skew scanned document images. Experimental results show that this method can align the images correctly and the efficiency of the method is better than traditional method. And it is robust to noise and other disturbance.