Abstract:
Reducing dimensions is a key step to improve the efficiency of face recognition.We have made some improvement bases on gabor wavelet and LDA technology.After reducing dimensions of wavelet,we will deal with the data towards two directions.The first one is to classify the data using LDA,the second is that after reducing dimension of PCA,then extract some geometrical characteristics eigenvectors.The eigenvectors mainly include the four corners of the eyes,apex nasi and two corners of the month.And this operation reduces the dimension again.Then based on the different effects of different data in the process of recognition,we assign them different weight,and then get the face recognition result.Contrast to the experiment,this method has better recognition rate.