利用主元分析(principal components analysis,PCA)方法作为评测手段分别在不同的虹膜图库中进行了实验,结果证明,选取靠近瞳孔的虹膜内环区域与整个虹膜区域的比率在大于37.5%时,或者暴露的虹膜两侧扇形的角度大于70°时,就可以获得区分个体的特征量,达到比较稳定的识别率。实验结果还给出了虹膜内环区域面积与识别率的对应关系曲线以及虹膜两侧扇形区域面积与识别率之间的对应关系曲线。
We investigate the performance of using partial iris for recognition by principal components analysis method in different iris image databases. The experiment results show that when the inner ring of the iris is over 37. 5% of iris pattern exposure,or the angle of the exposed sector is over 70 degree,the enough information can be obtained in human verification. The results also show the relationship between the percentage of the iris used and the accuracy rate.