An Improved Least Trimmed Square Hausdorff Distance Finger Vein Recognition

Abstract

In order to solve the problem of poor accuracy and matching speed in finger vein recognition, an improved Least Trimmed Square Hausdorff Distance (LTS-HD) finger vein recognition based on weighted matching point algorithm is proposed in this paper. The incremental neighborhood search method is used to achieve matching acceleration for Least Trimmed Square Hausdorff Distance algorithm firstly, and the optimal matching weights for different types of matching points is searched by Particle Swarm Optimization, at the same time, the particle box exclusion mechanism is applied to avoid premature convergence so as to achieve global particle optimization. On this basis, the directional matching points which can effectively characterize the vein pattern information are extracted, and the optimal weights of vein matching point is calculated after introducing the weights and model optimization. At last, the improved Least Trimmed Square Hausdorff Distance algorithm is used to achieve finger vein recognition by introducing the optimal weights. Compared with other algorithms, the results show that the proposed algorithm has a significant improvement in the objective indicators such as matching speed and accuracy.