基于小波包分解的虹膜图像质量评价算法  

Iris image quality evaluation method based on wavelet packet decomposition

作　　者：陈戟[1] 胡广书[1] 徐进[1] 

机构地区：[1]清华大学电机工程与应用电子技术系,北京100084

出　　处：《清华大学学报（自然科学版）》2003年第3期377-380,共4页Journal of Tsinghua University(Science and Technology)

摘　　要：在虹膜识别系统中,虹膜图像的质量是影响识别效果的重要因素。为了保证识别的准确性,必须选择满足识别算法需要的图像来进行特征提取和模式识别。在对现有虹膜图像质量评价算法分析的基础上,该文提出了一种新的基于小波包分解的算法,根据虹膜纹理的特点确定小波包的分解方式,选取纹理高频分量分布最集中的子频带作为特征子频带,以其能量作为判别图像质量的准则。对比实验结果表明,该方法可以快速、准确地判断虹膜图像的质量,实现了对图像的实时处理,结果合乎人眼的主观度量,且区分不同质量图像的效果优于其它评价算法。The iris image quality is an important factor influencing recognition effectiveness in iris recognition systems. For iris recognition accuracy, the image must enable effective character extraction and pattern recognition. An iris image quality evaluation algorithm based on wavelet packet decomposition was developed whose decomposition mode could be chosen based on the character of the iris texture with the image quality evaluated using the energy of the characteristic subband. The realtime algorithm can quickly and accurately evaluate the quality of iris images. The results based on the judgment of the human eye differentiate better than other algorithms.

关 键 词：虹膜图像 质量评估 小波包 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]