Online Palmprint Recognition Research Using Manifold Learning

Palmprint is an important biological characteristic with many advantages such as simple sampling, one and only feature, information in abundance which of important research value. In this paper, an image acquisition system based on scanner is developed. A new rotating correction algorithm and a new ROI extraction algorithm based on Vertical Internal Max Square(VIMS) are proposed. In the feature extraction model, feature extraction based on Locality Preserving Projections (LPP) and Block Wavelet Energy Coding (BWEC) are well developed. Then a new classifier designing method based on grading search is proposed. Finally, a practical recognition software is developed. Many experimental results proved the effectiveness of the algorithm and software. The main research is as follows:1. An image acquisition system based on scanner is developed based on the current research results. New ideas of rotating correction are provided by a new rotating correction algorithm and VIMS.2. Reduce the dimension nonlinearly of local palmprint images using LPP algorithm. Then the performance is deeply researched by experiments. Also, feature extraction based on BWEC is proposed and effectiveness is proved through experiments.3. A new classifier designing method based on grading search is proposed, which can overcome the disadvantages of single feature extraction method by combine the two methods organically. Effectiveness is proved through experiments.4. Practical recognition software is developed according to the experimental results above. Reliability is proved by testing and beneficial exploration is made by developing this software.