燕山大学学报2024，Vol.48Issue(3)：236-243,8.DOI:10.3969/j.issn.1007-791X.2024.03.006
基于面部视频的低运算量心跳检测
Low-Computation heartbeat detection based on facial video
摘要
Abstract
Heartbeat detection based on facial videos can enhance the credibility of facial recognition technology,enabling applications in military domains such as countering facial disguises,distinguishing between real and fake targets,and identifying forged videos.However,existing video-based heartbeat detection algorithms suffer from high computational complexity,limiting their real-time applicability under low computational capabilities.In this paper,the algorithm principles and the time delay of key steps are analyzed,and two low-complexity algorithms based on interpolation are proposed:the RGB-FFT algorithm and the RGB-ICA-FFT algorithm.Through testing,it is found that the interpolated RGB-FFT algorithm has the lowest computational complexity at a frame rate of 5 fps.The interpolated RGB-ICA-FFT algorithm achieves comparable accuracy to the RGB-ICA-FFT algorithm at a frame rate of 15 fps,but with a 61％reduction in computational complexity.关键词
心跳检测/人脸识别/独立成分分析Key words
heartbeat detection/face recognition/independent component analysis分类
信息技术与安全科学引用本文复制引用
丁国栋,鞠明,于坤灿,向新..基于面部视频的低运算量心跳检测[J].燕山大学学报,2024,48(3):236-243,8.基金项目
陕西省自然科学基金资助项目(2021JM-220) （2021JM-220）
