英文摘要 | The development of biometrics technology based on the prosperity of research activities of modern electrical integrated manufacture, fast and robust algorithms. The technology of fingerprint identification, a most promising branch in biomet- rics, has made a great process and been being widely applied in a variety of situations of personal identity verification and recognition. Fingerprint identifica- tion can be divided into three periods: preprocessing of fingerprint images, feature extraction and feature matching. In this thesis, we focused on the research of al- gorithms in these three periods and proposed our research opinions. At the same time, we designed a fingerprint verification system. In this thesis, we made a primary study on the fingerprint identification system, which involves a lot of basic problems in image processing and pattern recognition, e.g. image enhancement, image Fourier transform, image wavelet transform, information theory, feature extraction, feature matching, etc. The main contributions of this thesis include following issues: 1. We propose a modified Gabor filter design scheme for special fingerprint image enhancement. In this scheme, the filer is anisotropic and selection of filter coefficients is image adaptive. 2. For the applied images, we developed an oriented median filter and an ori- ented Gaussian filter, and incorporated them into our automatic fingerprint identification system. The proposed method is simple, fast, and robust, which is very appropriate for the realtime application. 3. In this thesis, a feature extraction method based on gray scale is proposed. Through a variety of experiments~ we improved those traditional algorithms and experimental results showed that our proposed method could extract features more accurately. 4. Through our investigation to the property of global texture, we presented a new inherent feature by down sampling the image orientation field, which is very helpful to improve the feature matching accuracy. 5. We improved the traditional Hough transform-based feature matching method by incorporating the above new global feature into feature matching procedure. Therefore, by employing the new feature, we can achieve high matching accuracy without the procedure of fingerprint classification. 6. Collaborated with another colleague, we designed a fingerprint verification system. In a word, in this thesis, we have made a lot of fruitful attempts and significant progresses on our fingerprint identification system, especially fingerprint image preprocessing. |
修改评论