Research On 2D Face Feature Extraction And Recognition Algorithm In Folded Face Mode

Face recognition technology has broad application prospects in recent years because of its advantages such as being safer,more confidential,more convenient,less oblivious,better in anti-counterfeiting,less easy to forge,difficult to be stolen,and available anywhere and anytime.At present,the more successful linear discriminating methods are principal component analysis method,linear discriminant method,two-dimensional principal component analysis method,and two-dimensional linear discriminant method.However,the one-dimensional method needs to transform the image matrix into a one-dimensional image vector,which causes difficulties for subsequent feature extraction.The unreasonable use of the sample class information in the two-dimensional method and the too high dimension of the extracted feature vector affect the classification speed.And these algorithms extract the global features of the image,and are easily affected by light and other factors.In order to solve the above problems,in the face recognition stage,the face is replaced by the folded face as the face recognition mode,and then the two-dimensional principal component analysis method and the linear discriminant analysis method are used for the recognition.First,after horizontally mirroring the face image,it is superimposed with the original imageand averaged to obtain the left-right symmetric face image.The half face of the image is taken as the new image sample;then the sample is firstly analyzed by two-dimensional principal component analysis.The feature extraction method is used to obtain the feature matrix,then feature extraction of the two-dimensional linear discriminant analysis method is performed on the feature matrix,a new feature matrix is obtained,and a new feature matrix is used to perform feature extraction on the sample;finally,the sample to be measured is compared with The training samples are processed in the same way to obtain the projection characteristic matrix,and the minimum distance between the training sample and the projection sample feature matrix of the sample to be tested is determined,and then the category of the sample to be measured is determined.In view of the above problems,the method of combining two-dimensional principal component analysis and two-dimensional linear discriminant analysis in the folded face mode is adopted.Compared with principal component analysis and linear discriminant analysis,this method not only solves the transformation from image matrix to image vector,but also adopts the folding face mode in the sample mode,which reduces the dimension of the sample and reduces the influence of lighting factors.At the same time,it makes up for the irrational use of category information in the two-dimensional algorithm and the disadvantage of the high dimension of the feature vector.This will not only improve the recognition rate but also reduce the training classification time.Experiments show that the total calculation time(training time plus classification time)is 95.43% after the introduction of the folded face mode.Compared with the separate two-dimensional principal component analysis,the recognition rate increased by 4.1% and the time decreased by 10.4%.Compared with the separate two-dimensional linear discriminant analysis,the recognition rate increased by 2.6% and the time decreased by 11.4%,compared with the complete face mode.The two-dimensional master analysis combined with two-dimensional linear discriminant analysis reduced the recognition rate by 1.1% over the period of time by 16.7%,and compared with the full face mode,the folding face pattern decreased by 51.08% in the matrix dimension.