The Research On The Development Of Fingerprint Identification System And Image Illumination Adjustment For Computer Vision

Biometrics refers to the automatic recognition of individuals based on their physiological and/or behavioral characteristics. Some of the physiological and/or behavioral characteristics that are inherent to the person and can be used for biometric recognition include fingerprint, face, voice, iris, palm print, hand-geometry, DNA and so on. With the appearance and popularization of e-bank, e-commence and kinds of intelligent cards, using biometric technology accurately and reliably for individual's identification have drawn more researchers' attention in pattern recognition and will have great potential market and broad application.In this thesis, the theory and technique of fingerprint recognition, image illumination adjustment for computer vision are investigated. A whole automated fingerprint identification system is developed. The dissertation's main content is organized as follows.In the part of fingerprint enhancement, the methods of the fingerprint enhancement based on Gabor filter which includes image segmentation, acquiring the orientation and frequency information, binarization and so on are studied.In the part of fingerprint feature extraction, the methods of image thinning, the image repairing (short ridge, burr, bridge, pore and so on), feature extraction and fake feature filtration are investigated and implemented.In the part of fingerprint matching, a new dot pattern fingerprint matching method based on gradual alignment and weighted matching score is proposed and implemented. Experiments on four databases of FVC2002 show that this fingerprint matching technique is more accurate and efficient compared with the conventional approach.An AFIS based on system requirement and functional analysis which includes fingerprint enrollment process, fingerprint verification process (the matching is 1:1), fingerprint identification process (the matching is 1: N) is designed and developed. The system's performance is tested in addition.In the system of computer vision recognition, various illuminated condition have great impact on object's detection and recognition. In this thesis, the conventional image illumination adjustment methods are studied and a new illumination adjustment algorithm based on statistical model is proposed and implemented. The experimental results test on CAS-PEAL face database show that the technique can efficiently improve the image lighting quality and thus reduce the negative effect of face detection and recognition under variable lighting.