基于Curvelet变换的虹膜识别

为提高虹膜识别的准确度,提出采用第二代曲波变换提取虹膜图像纹理特征的方法.首先将预处理后的虹膜图像进行曲波分解,然后采取不同方法对曲波分解后的低通和带通子带进行特征提取,并根据各带通子带的能量大小赋予不同的加权系数,最后采用欧式距离进行虹膜特征匹配.在4种虹膜图像库上测试表明,提出的虹膜图像纹理特征提取方法具有较高的识别率,与Harr小波和DCT(离散余弦变换)相比,特征提取时间相当,但识别率相对DCT来说,分别提高了1.8％,2.1％,2.0％和0.9％.

Iris Recognition Based on the Second Generation Curvelet Transform

In order to improve the accuracy of iris recognition,an iris image texture feature extraction method based on the second generation curvelet transform is put forward in this paper.Firstly,the preprocessed iris image is decomposed by this curvelet,then different texture feature extraction methods are used to lowpass subband and bandpass subbands,and giving different weight to each bandpass subbands according to their energy.Finally,Euclidean distance is adopted for iris feature matching.Experimental results on four iris databases demonstrate that the proposed method has better recognition rate.Compared with Harr wavelet and discrete cosine transform,they have the same time of feature extraction,but its recognition rate of the proposed method improved 1.8%,2.1%,2.0% and 0.9% respectively when it compared to DCT.