摘要
提出一种新颖的虹膜识别算法。该算法利用二维Log-Gabor滤波器对归一化后的虹膜图像进行滤波,滤波后的中高频子带图像被分割为若干子块,提取各子块局部极大值点并对其进行相位粗量化编码来生成虹膜特征码;特征匹配时使用虹膜特征码和干扰掩模来生成汉明距离,以此衡量两个虹膜之间的相似程度。该方法具有平移、放缩和旋转不变性,并且生成的虹膜特征码仅288B。实验结果和数据分析表明该方法的合理性和可行性。
A novel iris recognition algorithm is proposed in this paper. This method use 2D Log-Gabor filters to filter the normalized iris image, feature codes is generated through phase coding with rough quantization for every maximum point of each segmented blocks of sub-band image of middle and high frequency. In the moment of matching, hamming distance used to be the measurement of similarity between irises is calculated by feature codes and noisy mask. Our method are capable of great translation, scale and rotation invariant, the size of the feature codes is only 288 byte. Experimental result and data analysis show that our method is rational and feasible.
出处
《电子测量技术》
2007年第9期12-15,共4页
Electronic Measurement Technology
基金
河北省教育厅自然科学项目(2004124)
