具有活体检测功能的手背静脉身份识别方法研究 
下载： 1037次
Recognition Method of Dorsal Hand Vein with Liveness Detection Function
上海交通大学电子信息与电气工程学院仪器科学与工程系, 上海 200240
图 & 表
图 1. 由自主搭建装置拍摄得到的不同个体的静脉近红外图像
Fig. 1. Vein near infrared images of different individuals captured by self-developed setup
下载图片 查看原文
图 2. 活体检测的过程与结果。(a)(b)活体与假体的脉搏信号经过基线校正后的结果;(c)(d)去除直流分量后的活体与假体的脉搏信号;(e)(f)活体与假体脉搏信号的离散傅里叶变换结果
Fig. 2. Processes and results of liveness detection. (a)(b) Results of baseline correction of the pulse signals of the living body and the prosthesis; (c)(d) pulse signals of the living body and the prosthesis after removing the direct current component; (e)(f) discrete Fourier transform results of pulse signals of the living body and the prosthesis
下载图片 查看原文
图 3. 图像预处理的结果。(a)经过旋转平移位置校正的结果;(b)获取的ROI图像;(c)经过CLAHE增强的图像
Fig. 3. Results of image preprocessing. (a) Result of position correction after rotation and translation; (b) ROI image acquisition; (c) CLAHE for image enhancement
下载图片 查看原文
图 4. 样本的马氏距离
Fig. 4. Mahalanobis distance of samples
下载图片 查看原文
图 5. 决策树个数与错误率的关系
Fig. 5. Relationship between decision tree numbers and error rate
下载图片 查看原文
图 6. 随机森林模型的识别结果。(a)未经过预处理的识别结果;(b)经过预处理的识别结果
Fig. 6. Recognition results of random forest model. (a) Recognition results without preprocessing; (b) recognition results after preprocessing
下载图片 查看原文
图 7. 网格搜索法优化支持向量机参数c和g
Fig. 7. Support vector machine parameters c and g optimized by grid search method
下载图片 查看原文
图 8. 支持向量机模型的识别结果。(a)未经过预处理的识别结果;(b)经过预处理的识别结果
Fig. 8. Recognition results of SVM model. (a) Recognition results without preprocessing; (b) recognition results after preprocessing
下载图片 查看原文
表 1PCA的前150个成分的累积贡献率
Table1. Cumulative contribution rate of the first 150 components of PCA
| Principal component | Cumulative contribution rate /% |
|---|
| PCA1 | 21.34 | | PCA2 | 34.72 | | ︙ | ︙ | | PCA150 | 85.83 |
|
查看原文
表 2本文方法的实验结果比较
Table2. Comparison of experimental results of proposed method
| Method | Before PCA+MD | After PCA+MD |
|---|
| Accuracy /% | Error /% | Time /s | Accuracy /% | Error /% | Time /s |
|---|
| SVM | 99.42 | 0.58 | 0.156 | 99.86 | 0.14 | 0.110 | | RF | 98.27 | 1.73 | 1.919 | 99.28 | 0.72 | 0.368 |
|
查看原文
表 3本文方法与相关研究方法的结果比较
Table3. Comparison of results of proposed method and related methods
| Method | Accuracy /% | Time /s |
|---|
| Proposed method | PCA+MD+PLBP+RF | 99.28 | 0.368 | | MD+PLBP+SVM | 99.86 | 0.110 | | LBP+SVM[12] | 99.33 | 1.253 | | PLBP+KNN+SVM[27] | 99.30 | - |
|
查看原文
陈秀莲, 黄梅珍, 富雨超. 具有活体检测功能的手背静脉身份识别方法研究[J]. 光学学报, 2021, 41(6): 0610002. Xiulian Chen, Meizhen Huang, Yuchao Fu. Recognition Method of Dorsal Hand Vein with Liveness Detection Function[J]. Acta Optica Sinica, 2021, 41(6): 0610002.
PDF全文
