Research Of An Adaptive Hand-based Personal Identification

Biometric personal identification is emerging as a powerful means for automaticallyrecognizing a person's identity, and is attracting much attention lately. The biometriccomputing-based approach is concerned with identifying a person based on a knownbiometric feature database by image processing and pattern recognition technology, and it is akind of pattern recognition which applies biometric and informatics technology, taking thebiometric technology as basement and information as means. As a new biometric feature,palm-print has received much attention from researchers, because it has the advantages ofuniqueness, stability, containing more information and easily captured. Above all, the furtherresearch of palm-print identification technology can bring with not only the greatdevelopment of authentication, control and management theory, but also the benefit to societyand economy.Combined with the previous work of other researchers, a discussion is elaborately madeon the key techniques concerning image capturing, pre-processing, feature extraction andmatching and some improved and creative approaches to these problems are put forward. Anadaptive and robust verification technology by making use of hand-print features is proposed.Two hand-based features, the finger-print and the palm-print are considered. In contrast to theexisting approaches, our system employs both 1D and 2D hand-print signal to achieveeffective personal identification. The system consists of two parts: a convenient device forhand-print image acquisition and an efficient algorithm for fast palm-print recognition. Arobust and adaptive image coordinate system is defined to facilitate feature extraction. And adiscrete wavelet zero-crossing and 2D Gabor encoding scheme is applied to hand-printmulti-feature extraction and representation. In addition, a dynamic selection scheme isproposed to facilitate the fast search for the best matching of the sample in database in ahierarchical fashion. Particularly, wavelet zero-crossing coded 1D finger-print feature isapplied in the coarse level classification, and both 1D wavelet zero-crossing and 2D Gaborfeature are provided for the fine level matching. Based on the coding scheme and hierarchicalmatching strategy, this paper implements a hand-based personal identification system, andexperimental results demonstrate the effectiveness of the proposed system.