Face Recognition System Design And Algorithm

The critical problem of face recognition technology is the feature extractionalgorithm of the human face, that is, the key feature of an individual face is to beextracted, which is also the core point and necessary premise for matching the testsamples and training samples in the face recognition technology. Compared with otheridentity recognition methods, face recognition method is convenience, friendly, and fast,which is also the reason why it can be applied more and more widely and accepted bymore and more users.In this thesis, based on the face recognition technology, according to therequirement for staff registering attendance, a registering system for face recognition isdesigned. It stored the staff’s face data information first, including the face image, thename, and so on. The system can realize the identity recognition function for the staffsexisting in memory and the warning function for the illegal landing person. The purposeof face recognition system is to realize the recognition correctly, which can be dividedinto two phases, the first phase is the image acquisition and recording to build the basicdata platform of face image, the second phase is the recognition of the training samplesand test samples, which is the main and important part in this paper.Principal component analysis (PCA) algorithm is a most common technology in allface recognition feature extraction algorithms, which plays a very important andeffective role in statistical techniques of data analysis. The purpose of PCA is toproject high dimensional space vectors to low dimensional space vectors, which didnot change the main data and parameters of original data structure, so it can achievethe purpose for processing information conveniently and saving the data storagecapacity. BP neural network is composed of input layer, hidden layer and output layer,which has powerful operability. After using BP neural networks algorithm into themodel, the speed and the accuracy of face recognition are greatly improved. In thispaper, the identity recognition is realized according to the PCA algorithm and BP neural network algorithm together.