The application of scale invariant feature transform fused with shape model in the human face recognition

In order to enhance the facial recognition rate of the SIFT, the improved SIFT fused with shape model was proposed and applied to face recognition. Firstly the shape model was established by utilizing the training process of AAM model, and was used to get the initial positions of facial feature points, then the feature descriptors of SIFT which have a nearest Euclidean distance with the initial positions were reserved as the local facial feature, the improved SIFT local feature was combined with PCA for getting a mixed feature vector, finally the SVM was used to complete face recognition. The experimental results showed that, the recognition rate has increased 4.6% than PCA-SIFT under 5 training sample of each person in ORL database, and under different training samples, the rate is higher than the other five methods. It can be seen that the improved method can effectively improve the recognition rate, and the method is robust and effective.