摘要
常规的人脸特征点识别技术易受到图像分辨率降解作用影响,导致特征点的欧氏距离校准误差较高。为此,文章提出了基于图像关键特征增强处理的人脸特征点识别技术,在图像关键特征增强处理的基础上,完成了对人脸特征点的定位;设计了人脸特征点识别跟踪算法,结合定位与跟踪结果,实现了特征点识别。试验结果表明,应用新识别技术后,不同特征点的欧氏距离校准误差均较低,说明新技术具有有效性。
Conventional facial feature point recognition techniques are susceptible to image resolution degradation,leading to high calibration errors in the Euclidean distance of feature points.Therefore,the article proposes a facial feature point recognition technology based on image key feature enhancement processing,which completes the localization of facial feature points on the basis of image key feature enhancement processing;Designed a facial feature point recognition and tracking algorithm,combined with localization and tracking results,to achieve feature point recognition.The experimental results show that after applying the new recognition technology,the Euclidean distance calibration errors of different feature points are relatively low,indicating the effectiveness of the new technology.
作者
谭振华
王龙
TAN Zhenhua;WANG Long(School of Information Engincering,Jinzhong College of Information,Jinzhong 030600,China;School of Data Science,Jinzhong College of Information,Jinzhong 030600,China)
出处
《无线互联科技》
2024年第19期93-95,共3页
Wireless Internet Science and Technology
基金
基于图像分析的人脸识别对比技术研究,项目编号:2022L662
项目名称:基于跨“五维”多层次智能大数据自主创新实验室,项目编号:2022P018。