多算法融合虹膜识别中的规范化模型比较研究

Research on Normalization Model Comparison in Iris Recognition Based on Multi-Algorithmic Fusion

虹膜识别易受环境影响,利用多算法融合识别提高复杂应用环境下虹膜识别可靠性是一种非常有效的途径。本文针对多算法融合虹膜识别中的关键步骤——规范化模型选择做了比较性研究。首先搭建多算法融合虹膜识别平台,对常见的三种规范化模型在UBIRIS虹膜库中做了比较测试,实验结果证明双sigmoid函数指数模型性能最优。本文研究可对多算法融合的研究提供理论参考。

Iris recognition is susceptible to the environment, and multi-algorithmic fusion is an effective way to improve the performance of iris recognition in complicated environment. This paper makes a research on normalization model comparison. Iris recognition framework based on multi-algorithmic fusion is first built and three common normalization models are compared in UBIRIS iris database. Experimental results show that exponential model using sigmoid function can get best recognition performance. This work can give theoretical reference for the research of multi-algorithmic fusion.