基于平行双目立体视觉的三维指纹重建的研究-检测技术与自动化装置专业论文.docxVIP

西南科技大学硕士研究生学位论文第 西南科技大学硕士研究生学位论文 第 PAGE II 页 西南科技大学硕士研究生学位论文第 西南科技大学硕士研究生学位论文 第 III 页 Abstract At present, recognizing the identity through fingerprint has been widely used. The contact method is mostly used to acquiring fingerprint images, but it has inevitable limitations, such as when the finger touches and presses the mirror, traces of fingerprint may be stolen, and the pressing deformation also can make the fingerprint image partial distortion etc. At the same time, information of the 2-D image is limited and more fingerprint information is needed in some special applications, so this paper carried out the research on the reconstruction of non-contact 3-D fingerprint. This paper improves the methods based on the binocular stereo vision and proposes a new method based on the parallel binocular stereo vision. It only uses two cameras to collect fingerprints so as to reduce the costs of devices, while placing the two cameras in parallel can ensure the same area of the two fingerprint images in the cameras larger than that collected by placing the two cameras with an angle, so it can reconstruct more 3-D fingerprint information. In addition, the existing 3-D reconstruction algorithms are mostly based on the fingerprint surface, their results are not conducive to the subsequent recognition. Therefore, a 3-D reconstruction algorithm based on fingerprint ridges is presented in the paper, its reconstruction result directly contains the minutiae of fingerprint images, lines and other related information, and it is more conducive to the study of the subsequent 3-D fingerprint identification. The subject completed the hardware and software design of fingerprint collection system based on parallel binocular stereo vision and then present a new fingerprint image matching algorithm, which is the optimal topology matching algorithm based on the topological structure of the center point and minutiae. Finally, it implemented 3-D reconstruction based on t