基于步态识别的远距离身份认证

Human Identification from a Distance Based on Gait Recognition

通过人走路的姿势实现对个人身份的远距离识别和认证是当前生物特征识别研究领域的一个研究热点。算法利用步态轮廓图像边界到重心的距离矢量对步态轮廓图像进行人体运动的静态形状描述,采用连续隐马尔可夫模型对人体运动时从一个动作到另一个动作的过渡进行动态描述。算法在CMU数据库上面进行实验取得了较高的正确识别率。

Human identification from a distance is a new biometrics recognition technology.This paper attempts to describe gait contour by using the distance between the contour centroid and the pixels on the edge,and to solve the problems of human walking dynamics description using HMM.By utilizing the algorithm in the paper,the experiments made on CMU database achieve comparatively high correction identification ratio.