Face Recognition Directly Based On Two Dimensions Image

In recent years, considerable achievements have been made in face recognition. In order to enhance the accuracy of the recognition, the main goal of our research, we deal with the problems as follow: feature extraction, classifier selection and distance testing.Feature extraction is one of the most fundamental problems. In face recognition, to extract the valid discriminating features plays the key role to solve the problems. PCA and fisher are both the most typical methods that are based on linear projection analysis and widely used. This text does a further research on theories and algorithms of linear projection analysis. Dr. Jian Yang advanced 2D PCA for feature extraction which got some improvement both in time-consuming and accuracy of the recognition. Moreover, we advance a new method of face recognition, in which we regard each single face images as a training sample, then we do the feature extraction as well as classification, and gain a better accuracy. Ultimately, based on 2D PCA and the analysis of the 2 dimension linear discriminating theory, we advance the improved 2DPCA, L-2DPCA and LR-2DPCA.In conclusion, we advance a new method in face recognition, a face recognition algorithm directly based on face images. Before the recognition of faces, we do not extract the features (under no feature extraction's circumstance) then select the appropriate classifier and distance testing formula to categorize the testing images. Plenty experiment results indicate that this method can achieve higher accuracy, but keep in similar time complexity as 2DPCA.