Face Recognition Based on Random Sample and SRC

Recently classification based on sparse representation (SRC) has been successfully used in face recognition. SRC learns discriminative features from training set, and then classifies test samples by the minimum representation error. However, the dimension of face images is typically very high and leads to high computation complexity and excessive memory space, which can be effectively avoided by firstly adopting the method of random sampling and wavelet transform to reduce the dimension of face images, and then utilizing SRC for classification. The experimental results show Experiments results show that the accuracy and speed of face recognition can be efficiently improved by our algorithm.