提出了一种基于近红外人脸和虹膜的匹配层融合算法.该算法对近红外人脸运用小波变换结合二维主成分分析的特征提取方法和欧氏距离比较的匹配方法,对虹膜运用局部信息统计的分块编码方法和汉明距离比较的匹配方法,并在匹配层采用支持向量机策略对匹配分值进行融合,从而运用融合后的匹配分值进行决策.在一定规模的多模态数据库中进行了算法的验证,融合结果显示：基于支持向量机的近红外人脸和虹膜匹配层的融合,能够对较高的虹膜识别的准确率进一步提升,增强了系统识别率及鲁棒性.

Based on the near-infrared human face and iris,a fusion algorithm in the matching level was proposed.In the proposed algorithm,face was processed using two-dimensional principal component analysis (2DPCA) method based on wavelet transform for feature extraction and using Euclidean distance matching method for comparison.Iris was processed using the block-encoding method based on statistic of local information for feature extraction and using hamming distance matching method for comparison,was fused the match score using support vector machine (SVM) strategy in the matching level,and the fused matching score was used to make decision.The fusion algorithm was applied in a multi-model database,and the experimental results show that the SVM fusion algorithm in matching level combines the advantages of the original biometric and even expresses a higher strength of the total recognition rate,which enhances the robustness of the multi-biometrics recognition system.