Font Size: a A A

Research And Application Of Face Recognition System

Posted on:2019-10-23Degree:MasterType:Thesis
Country:ChinaCandidate:S Q ChangFull Text:PDF
GTID:2428330545990113Subject:Control Science and Engineering
Abstract/Summary:PDF Full Text Request
With the deepening of modern disciplines,face detection and recognition have received unprecedented attention and have been applied in practical production.The existing detection and recognition algorithms have high requirements for the computer's hardware and computing power.In the application of the low configuration embedded system,the real-time performance needs to be strengthened.Aiming at this point,this paper studies the problems in the process of face detection and recognition.In face detection and recognition,the most important thing is to accurately find the uniqueness of each face and describe it accurately in mathematical way.However,in fact,the matrix dimension of face image is larger,and the amount of computation increases with the increase of the accuracy of the picture.For example,the face image matrix collected by the camera in this paper is 320*280(89600 pixels),and the corresponding covariance matrix will reach 89600*89600 about 7.5G,which leads to a huge amount of memory and huge amount of computing consumption,which makes the existing recognition algorithms run hard on embedded systems.So we see many practical applications of face images are smaller,for example,the ORL standard face library has a matrix dimension of only 112*92.However,the smaller dimension of the image may lead to a lack of detailed description of the details of the face and the loss of detail differences.In order to solve the above problems,this paper divides and compresses the face image,that is,first select the parts of the eyes,nose,mouth and chin,which contain the most important features,and remove the ordinary part of the face information.In order to make up for the missing face global information,then the whole face graph is reduced to a small face map and then added.In this way,the segmentation and compression face contains six parts,namely left,right eye,mouth,nose,chin and reduced whole face map.After that,the six parts of the six parts are extracted by the principal component analysis(PCA).The feature vectors are selected according to the size of the eigenvalues,and the eigenvectors corresponding to the first 90%of the total eigenvalues are selected.Then,the support vector machine(SVM)is selected for classification and recognition.
Keywords/Search Tags:Face detection, face recognition, principal component analysis, support vector machine, Face image segmentation
PDF Full Text Request
Related items