基于三维动态步态的身份识别方法仿真  

Based on the Three-dimensional Dynamic Simulation Method for Identifying the Identity of the Gait

作　　者：李昆仑[1] 汪滢[1] 熊婷[1] 

机构地区：[1]南昌大学科学技术学院信息学科部计算机系,南昌330029

出　　处：《科学技术与工程》2016年第6期76-80,共5页Science Technology and Engineering

基　　金：南昌大学科技学院基金(2014-JG-04)和南昌大学科技学院精品课程项目(2013JPKC020)资助

摘　　要：根据步态识别人身份的研究中,由于二维步态特征无法完全表示人体特有的动态步伐特征,导致识别受限。提出基于三维动态步态的身份识别方法,以连续步态图像帧为单位,通过立体视觉技术从二维图对不同采样时刻的运动人体三维步态轮廓信息进行提取,提取人体步态三维轮廓后,对其进行无关区域分割,获取和人体行走相关的步态特征。通过构建步态特征变换的几何模型,给出动态三维步态特征的转换过程,依据欧氏距离度量,根据结果实现身份识别。仿真实验结果表明,所提方法具有很高的识别精度和识别效率,性能较静态方法有较大的改善。According to the people as to the identity of the gait recognition, the two-dimensional gait features can＇t completely said steps of dynamic characteristics of human characteristic, to identify constraints. Identification method based on 3 d dynamic gait was proposed, using continuous gait image frames as wnits, with stereo vision technology from 2 d figure human movement of different sampling time three-dimensional gait contour information extraction, extraction of human gait after three-dimensional contours, unrelated to regional segmentation, obtain and human gait features related to walk. In constructing the geometric model of the gait feature transformation, dynamic three-dimensional gait feature conversion process is given, based on the Euclidean distance measurement, according to realize the identification results. The simulation results show that the proposed method has high recognition accuracy and efficiency, performance is a static method has great improvement.

关 键 词：三维图像 动态步态 身份识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]