多指标融合的指纹图像质量评测方法  

Fingerprint Image Quality Evaluation Method with Multi-index Integration

作　　者：刘莲花[1] 谭台哲[2] 

机构地区：[1]华南师范大学增城学院计算机系,广州511363 [2]广东工业大学计算机学院,广州510006

出　　处：《计算机工程》2011年第9期226-228,共3页Computer Engineering

基　　金：广东省自然科学基金资助项目(05006349)

摘　　要：提出一种将分级排除和综合评测相结合的多指标融合的指纹图像质量评测方法。该方法综合考虑指纹图像的局部特征和全局特征,通过从空域和频域上提取质量评测指标,采用非线性加权平均的数学模型计算指纹图像的质量得分,得到对指纹图像质量的客观评测结果。在FVC2004标准指纹库中进行实验,结果证明该方法可以准确、客观地评测指纹图像质量,评测结果与专家的视觉评测结果接近。This paper presents a fingerprint image quality evaluation method based on multi-index integration,which combines progressively excluding and comprehensive evaluation.It combines local features and global features,and extracts quality evaluation indexes from spatial and frequency domain.Quality indexes are extracted and the quality scores are calculated by non-linearly summation with different weights from different quality evaluation indexes.It is tested in standard database of FVC2004,and experimental results show that the method is reasonable and effective,which is closed to experts＇ evaluation.

关 键 词：指纹图像 质量评测指标 多指标融合 非线性加权平均 质量得分 

分 类 号：TN911.73[电子电信—通信与信息系统]