基于PCA的人脸识别opencv代码.docxVIP

基于PCA的人脸识别opencv代码 # inelude 〃StdAfx. H〃 include stdio, h inelude string? H include CV? H〃 # include 〃cvaux? H〃 # inelude highgui. H〃 H # inelude time. # include iostreanT Using the namespace CV; Using namespace STD. / / define several important global variables IplImage * * fdcelmgArr = 0; / / Pointers to training faces and test faces (points to different learning and recognition stages) CvMat * personNumTruthMat = 0; / / the ID of the face image Int nTrainFaces = 0; / / the number of training images Int nEigens 二 0; / / the number of major eigenvalues that you take Ip1Image * pAvgTrainlmg 二 0; / / the average of training face data Ip1Image * eigenVectArr = 0; / / projection matrix, which is the main eigenvector CvMat * eigenValMat 二 0; / / eigenvalue CvMat * projectedTrainFaceMat 二 0; / / the projection of the training image / / function prototype Void learn (); Void recognize (); Void doPCA (); Void storeTrainingData (); Int loadTrainingData (CvMat * pTrainPersonNumMat); A float * projectedTestFace; Int loadFacelmgArray (char * filename); Void printUsage (); / / the main function, which consists of learning and identifying two phases, needs to be run twice, separated by the parameters passed by the command line Int main () { / / learn (); Recognize (); } / / the learning phase code Void learn () { Cout / / start time Clock_t start and finish; Double duration; Start 二 clock (); Int I, offset; / / load training images NTrainFaces 二 loadFacelmgArray (“ train. TXT “); If (nTrainFaces 2) { Fprintf (stderr, Need 2 or more training faces \ n〃 Input file contains only % d \ rT, nTrainFaces); The return; } / / the main component analysis DoPCA (); / / the training atlas is projected into the subspace ProjectedTrainFaceMat 二 cvCreateMat (nTrainFaces, nEigens, CV_32FC1); Offset 二 projectedtrainfacemat- step/sizeof (float); For (I = 0; I nTrainFaces; I + +) / / int offset 二 I * nEigens? CvEigenDecomposite ( FacelmgArr [I]? NEigens, EigenVectArr, 0, 0, PAvgTrainlmg, /