摘要
针对新兴的手背静脉识别技术,提出了一种具有位移和旋转不变性的局部尺度不变特征(SIFT,scale invariant feature transform)分析方法。首先确定手背静脉图像的感兴趣区域(ROI)并对其进行滤波去噪,然后提取手背静脉血管的SIFT并对特征点进行匹配,最后计算注册样本和待识别样本的特征匹配率并以此作为相似性测度进行身份识别。利用我们建立的手背静脉血管图像数据库对该算法进行了性能测试,并与目前最典型的识别方法进行了对比。实验结果表明,本算法具有更好的识别性能,其中识别速度得到了很大的提高。
Aiming at the emerging hand vein pattern recognition technology,an algorithm with translation and rotation invariance is developed in this contribution, which is based on local scale invariant feature transform（SIFT） feature analysis. First, the region of interest （ROD of the hand vein image is obtained and filtered to reduce the image noises, and then SIFT features are extracted and matched. Finally the matching ratio of features between the registered image and test image is calculated as the similarity measurement to verify the human identification. The performance of the proposed method is evaluated in the hand vein image database constructed by our group, which is also compared with the classical recog- nition image matching algorithm. The results show that the presented approach has better recognition performance, especially the speed which has been improved greatly.
出处
《光电子．激光》
EI
CAS
CSCD
北大核心
2009年第5期681-684,共4页
Journal of Optoelectronics·Laser
基金
国家自然科学基金项目-仪器专项资助项目(60627002)
天津市应用基础重点资助项目(06YFJZJC00400
043800511)
