|
| 基于轻量型网络的唇纹识别算法研究 |
| Research on Lip Print Recognition Algorithm Based on Lightweight Network |
| |
| DOI：10.16018/j.cnki.cn32-1650/n.202204009 |
| 中文关键词: 深度学习 卷积神经网络 特征提取 唇纹识别 图像识别 |
| 英文关键词: deep learning convolutional neural network feature extraction lip-print recognition image identification |
| 基金项目: |
|
| 摘要点击次数: 32 |
| 全文下载次数: 49 |
| 中文摘要: |
| 在信息化和智能化高度发展的大数据时代，身份信息安全面临着种种挑战，传统的身份识别技术已不能满足公众安全需求。为解决传统唇纹识别算法中图像预处理过程复杂、特征提取困难和识别周期较长等问题，提出基于卷积神经网络的唇纹识别算法，搭建一个轻量型神经网络LNet-6(lightweight network-6)。该网络模型具有参数计算量少、模型文件小和可移植性强等优势。直接输入原始数据集，简化图像预处理步骤，通过卷积层自动提取特征信息和下采样操作降低模型训练参数，避免了图像特征提取算法的复杂设计。在测试集上获得了97. 97%的识别率，验证了该方法的有效性。 |
| 英文摘要: |
| In the era of big data with highly developed informatization and intelligence, identity information security is facing various challenges. Traditional identity identification technology can no longer meet the needs of public security. In order to solve the problems of complex image preprocessing process, difficult feature extraction and long recognition cycle in traditional lip print recognition algorithm, a lip print recognition algorithm based on convolutional neural network is proposed, and a lightweight neural network LNet-6 with 6 convolutional layers is built. The network model has the advantages of less parameter calculation, small model files and strong portability. The original data set can be directly put in and the image preprocessing steps can be simplified. It can help extract feature information through the convolutional layer automatically and reduce the model training parameters through the down-sampling operation in order to avoid the complex design of the image feature extraction algorithm. The recognition rate of 97. 97% was obtained on the test set, which verified the effectiveness of the method. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |