Font Size: a A A

Research Of Face Detection And Recognition Technology

Posted on:2013-05-16Degree:MasterType:Thesis
Country:ChinaCandidate:J TongFull Text:PDF
GTID:2248330395956453Subject:Optical Engineering
Abstract/Summary:PDF Full Text Request
Face recognition occurs as a new biometric identification technique with the rapiddevelopment of image processing, computer vision, pattern recognition and so on in recentyears. It mainly includes two aspects of face detection and face recognition. Compared toother biometrics, face recognition technique is simpler, more intuitive, and having morehidden capability. Therefore, this technology has broad application prospects and importantvalue of theoretical research in information security, criminal investigation, public utilitiesand other fields.In this paper, the three parts of face possessing as follows will be mainly introduced: thepart of face detection, the extraction part of facial feature, the part of classification for sampleface.1) In the part of face detection, we use the color space based on YCbCr for face detection.This algorithm is a feature-based face detection method; it is not limited to the shape and sizeof the face, and easy for programming. After the experimental verification, the algorithm hasgood test results for different background, facial size, and expression profile and so on.2) In order to improve the probability of face recognition, we study separately the facialfeature of image in subspace and frequency domain. The PCA algorithm based on the optimalunit orthogonal vector basis is used in subspace domain. Meanwhile, the DMFE algorithm forfeature extraction is analyzed in frequency domain.3) For the sake of determining the identity of the current face, we use the SVMclassification algorithm to the extracted feature vectors for classification in this paper.In order to test and verify the impact of different algorithms on the recognition rate, wechoose the forty facial images from the ORL face database, the first5per person of which asthe training sample, and the remaining samples for identification. The results show that: thewavelet algorithm recognition rate is96.5%, and PCA is82%. The results further indicate thatthe SUM has especially better performance in for small sample, high dimension, nonlinearaspects of face detection and recognition, which is based on the wavelet transformation.
Keywords/Search Tags:Face recognition, face detection, principal component analysis, Wavelet transform, support vector machine (SVM)
PDF Full Text Request
Related items