Abstract: Most fingerprint identification algorithm has low accuracy for incomplete image, a novel miss fingerprint identification based on linked features of geostatistics and subset matching is proposed in this paper. Image preprocessing is performed by Gabor filter, binary conversion and thinning. The image is partitioned into several sub-images without overlapped image. The features of each sub-image are extracted such as the linked features of geostatistics, fingerprint minutiaes of fork point and endpoint. The performance of algorithm is test by simulation experiments. The result shows that the proposed method has achieved higher identification accuracy in missing images and has not increased consuming time.

Key words: fingerprint identification, incomplete image, linked features, geostatistics

摘要： 针对主流指纹识别算法对缺失指纹图像识别率非常低的问题，提出了一种地统计关联特征与多子集匹配的算法（GS-MS）。首先对指纹图像进行Gabor滤波增强以及二值化、细化预处理，然后将图像均匀划分为N个子集，分别提取各子集的地统计学关联特征与分叉点、端点等细节特征点，最后以待识别指纹图像子集为基准，与指纹库子集进行匹配识别。采用完整与缺失两种指纹数据集进行测试，GS-MS算法均取得了较优的识别精度，而且没有大幅度增加运行时间。

关键词: 指纹识别, 缺失图像, 关联特征, 地统计学