Research On Face Recognition Using Bidirectional Optimization Of2DPCA Method

With the development of information technology, information security awarenessis growing. Applications such as e-commerce, finance, trade, network security and moreefficient authentication technology. As a pattern recognition areas, the rise of newbiometric technology by virtue of its unique advantages recognized researchers to builda tough safety net for information security under the new situation, as well as theidentity of the identification study opens a new door.The reason why the face recognition technology to stand out from the manybiometric identification technology, has become one of the hottest research topics in thefield of identification, because it has obvious advantages, these advantages are difficultto explore other identification issues or application. These objective reasons for therapid development of face recognition technology achievements, and also to bring thehighest point of the identification of research areas. Therefore, in recognition of theunique technological advantages and broad application context, and how to obtain moreefficient identification rate naturally become the focus of researchers discussed thecurrent stage.In the face recognition technology, efficient feature extraction is the most corework. Traditional principal component analysis (PCA) was required in the imagerecognition of the image matrix into a vector, resulting in the number of the dimensionof the image vector is on the high side, so the larger the amount of calculation for theentire process of feature extraction. On the basis of the PCA, it was suggested that thetwo-dimensional principal component analysis (2DPCA), but the nature of the imagematrix row feature extraction, eliminating the image column, but still ignores row.Therefore, on the basis of in-depth study of the classical principal component analysismethod, proposed an improved2DPCA method, this method can eliminate the imagerows, columns, and expect to get a higher recognition rate.Paper selected the multiple mainstream face database as an experimental databaseto verify the feasibility and effectiveness of the improved method of this article in theMATLAB experimental platform on, while the improved method and PCA,2DPCA classic method by line comparison, the experimental results show that the improvedmethod has a more excellent effect of recognition and has more robust to facialexpressions or light-induced changes in the face.