Research On Sampling And Recognition Technolog Of High Resolution Palmprint

Nowadays, Biometric technology is kind of an important personal identification technology and wins more and more applications. Many biometric technologies based on human biological characteristics are being developed quickly. Palmprint recognition has been widely used in various identification systems, due to its advantages of easily sampling, uniqueness, stability and invariance. In this paper, the palmprint identification technology is studied and analyzed deeply, with the throughout process of palmprint image acquisition, preprocessing, feature extraction and feature recognition.Now palmprint image sampling is mostly to capture the main characteristic of palmprint, and so how to get high resolution image is the main target in this paper. With the acquired feature information on the palm of hand, such as the papillary pattern, the main line and the fold lines, and the information fusion technique, a high accuracy recognition system is established.Palmprint sampling system based on the principle of optical total reflection is designed, which is a double telecentric system. The double telecentric optical system is constructed with a front single lens and back gauss lens. The photoelectric detector of the sampling system is a CCD with 8 million pixels. The result of the design shows that the optical transfer function is close to diffraction limit at the cutoff frequency of the CCD, the RMS of the spot diagram is within Airy spot and the maximum distortion is less than 0.5%.Lighting system designed is a kind of Kohler illumination system, with the light source of LED. The system consists of a pipe integration device and a middle transfer lens, and the irradiance on the hand is uniformly. In the case of 10×10 sampling points the non-uniformity of the illumination is 9.5%. The result is basically satisfied to the requirement of the image acquisition with high resolution and low distortion.For the feature recognition, multi-frequency analysis method is used in palmprint feature analysis because the palmprint has strong directional feature. The palmprint image is decomposed by wavelet in different direction, and in every wavelet the discrimination of the palm is studied. The palmprint recognition is finally fulfilled by the integration of the feature information of the palmprint image.