摘要
针对利用2D掌纹图像进行身份鉴别时,因图像采集过程中丢失了部分手掌结构的深度信息,致使传统掌纹识别方法识别率并不高的问题,提出应用分块ST法对2D掌纹图像进行特征提取。在提取特征前,首先对预处理掌纹图像得到的感兴趣区域(ROI)使用具有各向同性的高斯滤波器滤除高频噪声,然后把滤波后的ROI分成若干块,求取每一子图像块的ST特征量,所有子块ST特征量的组合构成该幅掌纹图像的特征向量,最后利用最近邻分类器进行分类。应用该算法在UST手形图像库上进行了测试,获得95.5%的识别率且识别时间满足应用要求。实验结果表明:基于高斯滤波和分块ST的2D掌纹识别能够满足考勤、门禁等一些小型生物特征识别系统的应用。
In order to solve the problem that the recognition rate of traditional palmprint recognition method is not high because of losing the partial depth information of palm structure during image acquisition when 2D palmprint images are used for identification,this paper proposes to apply blocked ST method to feature extraction of 2D palmprint image.Before the features are extracted,the region of interest(ROI)of palmprint image obtained by preprocessing is firstly filtered by isotropic Gaussian filter with filtering high-frequency noise.Then,the filtered ROI is divided into a certain number of blocks,and the ST feature values of each sub-image block are calculated.The combination of ST features of all the sub-image blocks constitutes the feature vector of the palmprint image.Finally,the nearest neighbor classifier is used for classification.The algorithm is tested on UST hand image library.95.5%recognition rate is obtained and recognition time satisfies the application requirements.
出处
《工业控制计算机》
2021年第4期57-59,共3页
Industrial Control Computer
关键词
生物特征识别
2D掌纹图像
高斯滤波
分块子图像
ST特征量
biometrics recognition
2D palmprint image
Gaussian filtering
blocked sub-image
ST feature vector
