基于多通道Gabor滤波和D-S证据理论的虹膜识别

Iris Recognition Based on Multi-channel Gabor Filtering and D-S Evidence Theory

为了提高虹膜识别的准确性和稳定性,研究了Gabor滤波的虹膜特征提取方法,并在此基础上对虹膜特征进行了D-S(Dempster-Shafer)证据理论的改进.提出了多通道Gabor滤波和D-S证据理论的虹膜识别方法.该方法充分考虑到虹膜图像获取中被考察对象的姿态和环境光照等不确定性因素对识别结果的影响,通过选择不同频率和方向的Gabor滤波器组,有效提取虹膜特征,并结合D-S证据理论实现了虹膜特征决策.实验结果表明,所提方法增强了虹膜识别过程中抵御图像噪声、干扰和亮度变化等不利因素的能力,与标准的Gabor方法相比,识别率平均提高了4.05%.

To improve the accuracy and stability of iris recognition, this paper studies a feature extraction method via Gabor filtering and modifies the iris features using D-S （Dempster-Shafer） evidence theory. Then, an iris recognition method based on muhichannel Gabor filtering and D-S evidence theory is developed. Taking into consideration the influences of uncertain factors on iris recognition such as poses of the observed subject and environmental illumination changes during capturing iris images, the proposed method extracts iris features effectively through employing banks of Gabor filter at different frequencies and directions, and makes decision on iris features combining with D-S evidence theory. Experimental results indicate that the proposed method improves the performance excellently for iris images with noise, disturbance and environmental illumination changes, and increases the accuracy in recognition rate at an average 4.05% more than the conventional Gabor method.