人体手背静脉血管图像增强与特征匹配  

Image enhancement and feature matching of palm-dorsa vein

作　　者：吕佩卓[1] 赖声礼[1] 

机构地区：[1]华南理工大学电子与信息学院,广东广州510640

出　　处：《计算机工程与设计》2007年第15期3623-3625,3635,共4页Computer Engineering and Design

基　　金：广东省科技攻关基金项目(2003C40406)

摘　　要：为提高手背静脉血管图像识别的准确性和改善图像采集的灵活性,提出了一种新的手背静脉图像识别算法。使用自适应直方图均衡法增强背景和静脉纹理图像的总体对比度和局部细节,使用分水岭算法得到静脉纹理图像的单像素级图像,最后采用纹理相似性原理进行特征匹配和识别。试验结果证明,该算法在图像采集位置灵活变动情况下的错误接受率为0%,错误拒绝率4.6%,利用此算法进行手背静脉血管图像识别是可行的,具有良好的应用前景。To improve the veracity ofpalm-dorsa vein pattern identification and the maneuverability ofimage collecting, a new algorithm about palm-dorsa vein image identification is proposed. Adaptive histogram balancing is used to enhance the contrast of the background and the vein texture image, and watershed transform is adopted to produce the single pixel image of vein texture and texture comparability theory is used to carry out image matching and identification, Experimental results show that the FAR and FRR are 0% and 12% respectively, and it is feasible to use the algorithm to carry out palm-dorsa vein pattern identification.

关 键 词：静脉识别 自适应直方均衡化 分水岭算法 特征提取与匹配 相似性原理 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]