A Hierarchical Online PalmPrint Recognition Research With Multiple Features

Automatic personal identification is a significant component of security systems with many challenges and practical applications. The Advances in biometric technology have led to the very rapid growth in identify authentication. The palmprint, as a relatively new Biometric technology, have advantages over other biometric technologies, and the use of palmprint for identity authentication has drawn research and development from some researchers at recently years. But the development of an effective and efficient approach to recognition palmprint remains a challenging research topic. We deeply discuss and research these problems in this paper.At first, based on deeply research the palmprint recognition technology, we discussed some key issues, such as: the palmprint image preprocessing, the feature extraction and representation, the similarity measurement, fast matching, etc. and comparing existing various algorithm at present, and giving out their advantage and shortage. In contrast to the existing systems that employ a fixed mechanism for feature extraction and similarity measurement, we propose a hierarchical multi-feature scheme to facilitate coarse-to-fine matching for efficient and effective palmprint recognition. In our approach, three-level feature are defined: global geometry-based key point distance (level-1 feature), filtering feature (level-2 feature), and line feature (level-3 feature). We extract multiple features and adopt different matching criteria at different levels. At last, a series of experiment have been test the feasibility and efficiency of our approach. We mainly carry on an investigation from two aspects: the recognition rate and verification rate. The experiment result show that the recognition rate making use of our method is 97. 33%, and the false rejection rate (FRR) and false acceptance rate (FAR) can attain lower level. These show our approach can satisfy a request of biometric technology.The main contributions of the thesis have: (1) Carried on thorough research and compared for the existing palmprint recognition technology, pointed out their merits and shortcoming; (2) Proposed a hierarchical multi-feature scheme to facilitate coarse-to-fine matching for efficient and effective palmprint recognition; (3) Put forward a new feature we called PalmCode based on the Real Gabor Filter; (4) Considering the palmprint lines is very irregular and very hardly accurate depicted with mathematics, we proposed an approach that approximately represents these palmprint lines with some line segments, and gives its algorithm; (5) Put forward employing the line Hausdorff distance method to match two sets of line segments, and improved this method.