摘要
在视频侦查中,步态的检验主要依赖个人的观察能力和经验,存在主观性,导致对同一个体步态特征的判断存在差异。有必要提出基于视频的步态检验方法的构想,用于辅助侦查。采用文献分析、比较分析、调查研究的方法,通过视频帧序列提取2D关节点,映射3D姿态序列,参照三维动捕设备得出的步态特征参数,以实现视频图像中步态特征参数精准测量,再运用似然比的方式评估步态检验特征。提出步态侦查今后的一个发展方向,为提高侦查效率、步态在法庭上的证据价值提供了研究基础。
In video investigatio,gait inspection mainly relies on individual observation ability and experience,which is subjective and leads to differences in the judgment of gait characteristics of the same individual.Therefore,it is necessary to propose the conception of a video based gait inspection method to assist in reconnaissance.Literature analysis,comparative analysis,and investigation and research methods are utilized,2D joint points are extracted from video frame sequences,and 3D pose sequences are mapped.Then,gait feature parameters obtained from 3D motion capture equipment are referenced to achieve accurate measurement of gait feature parameters in video images.Finally,likelihood ratio method is used to evaluate gait inspection features.The method has been demonstrated by experts and has strong theoretical innovation value.It proposes a development direction for gait investigation in the future,providing a research basis for improving investigation efficiency and the evidential value of gait in forensic science.
作者
冯磊
FENG Lei(School of Criminal Investigation,People’s Public Security University of China,Beijing 100038,China;Institute of Forensic Science,Ministry of Public Security,Beijing 100038,China)
出处
《指挥与控制学报》
CSCD
北大核心
2024年第3期372-380,共9页
Journal of Command and Control
基金
中央级公益性科研院所基本科研业务费专项(2023JB012)资助。
关键词
步态检验
二维视频
三维步态
人工智能
侦查应用
gait inspection
two-dimensional video
three-dimensional gait
artificial intelligence
investigation applications