高级检索

面向虹膜识别的多方向中心对称局部二值模式

Iris Texture Characterization of Multi-direction and Center-symmetry Local Binary Pattern

  • 摘要: 为提升虹膜纹理的表征效果, 获得稳定的局部纹理信息表达, 提出基于多方向中心对称局部二值模式的虹膜识别算法. 首先通过眼睑边缘快速定位生成掩膜, 抑制眼睑区域噪声; 然后在多方向的基础上提出多方向中心对称的局部二值模式完成虹膜纹理特征的表征; 最后利用汉明距离来判定虹膜是否匹配. 针对由于类内与类间匹配次数失衡导致评价指标正确识别率存在的不足, 提出修正的正确识别率. 在CASIA.V1, CASIA.V3-Interval, JUL6.0和CASIA.V4-Lamp库上取得的修正正确识别率分别为99.857%, 99.940%, 99.640%和97.973%, 等错误率下降了0.03%~5.96%, 实验结果表明所提算法能够有效地表征虹膜纹理信息, 具有良好的识别性能和鲁棒性.

     

    Abstract: In order to improve the iris texture characterization effect and obtain stable expression of local texture information, an iris texture characterization method based on multi-direction and center-symmetry local binary pattern is proposed. Firstly, a mask generated by quickly locating the edge of the eyelid is used to suppress the noise in the eyelid area; secondly on the basis of multi-direction, a proposed center-symmetry local binary pattern is used to characterize the iris texture; finally, the shifting Hamming distance is used to determine whether the iris is matched or not. And specially to be mentioned is that the traditional correct recognition rate (CRR) is modified to deal with the shortcomings of the CRR evaluation index due to the imbalance of intra-class and inter-class matching times. The modified correct recognition rates obtained on CASIA.V1, CASIA.V3-interval, JUL6.0 and CASIA.V4-lamp libraries are 99.857%, 99.940%, 99.640% and 97.973%, respectively, and the equal error rate decreased by 0.03%~5.96%. The experimental results show that the proposed algorithm can effectively characterize the iris texture information, and has good recognition performance and robustness.

     

/

返回文章
返回