基于二维正交滤波的虹膜识别算法

提出一种针对灰度虹膜图像进行虹膜特征提取及匹配的方法,其利用四元数二维正交Log Gabor小波提取虹膜图像的纹理特征,以滤波后图像的解析信号作为虹膜的特征编码.该方法可以同时表征虹膜纹理多方向上的特征,更加全面地描述了虹膜纹理的特征空间.特征匹配采用类似汉明距的方式,同时以虹膜图像中眼睑、睫毛以及光斑的分布为匹配模版来减少它们的干扰.大量实验的结果表明该方法具有非常优越的识别性能.

Reliable Iris Recognition Using 2D Quadrature Filters

This paper investigated a new approach for iris recognition with gray iris image.The approach extracts iris feature using 2D quadrature Log Gabor filter obtained by quaternion fourier transform.The method can get reliable iris phase information by orthogonal directions and calculate analytic signal of the filtered iris image as feature.The method characterizes iris feature in multidirectional ways.Thus iris feature encoding with high complexity that makes the iris feature more comprehensive is obtained.Feature matching adds template and uses the similar way of hamming distance to reduce the disturbance of eyelid,eyelash and facula.The results with 2D Log Gabor quadrature filters achieve significant recognition performance.