Font Size: a A A

The Safety Research For Face Images Based On Digital Watermarking

Posted on:2014-01-20Degree:MasterType:Thesis
Country:ChinaCandidate:Y Y WangFull Text:PDF
GTID:2248330395489600Subject:Measuring and Testing Technology and Instruments
Abstract/Summary:PDF Full Text Request
Today, biometric technology has attracted more and more attention, and face recognitionwith its own unique characteristics, as one of its main technology, has been widely applied.However, malicious attacks to face images’ become more easily in the pace with the rapiddevelopment of computer network communication technology. So face recognition bringsus the facilitate service as well as the challenge. At present, an increasingly attention isattracted to this new field of digital watermarking and this brings digital watermarkingtechnology a more rapid development.To dissolve the problem that the misrecognition or wrong identification caused by faceimages’ malicious attack. Two digital watermarking methods(fragile digital watermarkingof face images based on wavelet transform and LSB) has been presented, after the analysesand research of digital watermarking and face recognition technology. In these twomethods, we first embedded the designed watermark information into the face images,second extracted the watermark information to detect malicious temper to images andlocate the attack position, and last tested using the sample face images of ORL database.Experimental results show that the two methods all be with the good invisibility, detectionand location abilities to malicious attacks, and has no effect on the face recognition rate.A system of face recognition was established in this paper. First is the PCA method’suse of feature extracting to the face images. Most of the energy is concentrated in the firstfew base functions with larger Eigen values (having larger amplitude), and less energyconcentrated in the basis functions whose coefficient is with smaller magnitude. Euclideandistance method is employed to match the test face image and the training images. And lastimplementing by programming, the algorithm has a high face recognition rate testing bythe sample face images of ORL database.
Keywords/Search Tags:Face recognition, fragile digital watermarking, discrete wavelet transform, LSB
PDF Full Text Request
Related items