Font Size: a A A

Palmprint Recognition Based On Two-dimension Gabor Filter

Posted on:2008-07-20Degree:MasterType:Thesis
Country:ChinaCandidate:H J WuFull Text:PDF
GTID:2178360212996841Subject:Pattern Recognition and Intelligent Systems
Abstract/Summary:PDF Full Text Request
With the rapid growth of e-commerce in the modern society, biometric technologies have been widely used in personal identification. Biometrics-based recognition is a personal recognition technology that combines information technology and biology technology to recognize a person. Personal characteristics including physiological and behavioral characteristics are universal, unique and permanent, so they can be used to reinforce or even replace the traditional authentication tools. Biometrics technology can be widely used in national security system, financial system, social welfare system.PaImprint, one of the most important biometric features, also has important application value in many fields. Palmprint is the inner surface of a hand between wrist and fingers. As an important complement of the biometric features,palmprint has its unique characteristics. Palmprints contain more distinctive information than fingerprints, and are more stable than signatures. Palmprint based recognition can obtain higher accuracy than face recognition. And palmprints contain distinctive features such as principal lines and wrinkles, which can be extracted from low-resolution images.This paper mainly researches on palmprint identification's key technology, especially the feature extraction and matching. And a digital camera-based palmprint recognition system has been developed. And the follow are the work of this paper.1 . We design an easy image capture device. Before feature extraction, we need to preprocess the images we get. We adopt the easy way to locate the palmprint and get a good result. And we program to realise the overall processes - makegray, binarize, contour extraction, localization and get sub area, etc.2. Because two- dimension Gabor function is locally optimal in both time domain and frequency domain, and it has excellent selection feature about orientation and frequency, the paper proposes a filter based on two- dimension Gabor function to extract the texture feature of the palmprint. And we adopt FFT to accelerate the rate of convolution. We also make a lot experiments to choose the parameter of the Gabor filter.Use the adjusted Gabor filter to convolute with the centre palm, which is obtained by the pre-processed step. And then code to form the real part matrix and the imagine part matrix. In this way, the texture feature of the centre palm has been obtained.3. In the stage of Identification, we use the normalized Hamming distance to describe the similarity of the two palms. We use the palms which are partially obtained by ourselves and the others are the palms of The Hong Kong Polytechnic University .We use the threshold we set to do test the experiment. When we set the threshold 0.335, the FRR (False Rejection Rate)and FAR(False Accept Rate) are 0.9% and 0% respectively to the palms of The Hong Kong Polytechnic University .While to the palms which are obtained by ourselves, the FRR and FAR are 0.85% and 0.12% respectively, and this result is not so good. But when we set the threshold 0.325, the FRR and FAR are 1.68% and 0%. We also make a summarize of the technique we suggest, and the result is encouraging.We use vc++6.0 to realize the experiment.The paper have extracted the texture feature of the palmprint using 2-D Gabor filter, the result of matching identification is encouraging. Giving us some affirmation of the experiment.
Keywords/Search Tags:Biometrics, Palmprint recognition, Pre-processing, Feature extraction, Gabor filter, Identification
PDF Full Text Request
Related items