步态识别是一种新兴的生物特征识别技术,旨在通过人们走路的姿态进行身份识别。与其他的生物识别技术相比,步态识别具有非接触远距离和不容易伪装的优点。提出了一种基于新的特征提取方法的自动步态识别算法,该算法仅从腿部的运动进行身份识别。对于每个序列,用一种基于图像色度偏差的背景减除算法来检测运动对象,在经过后处理的二值图像序列中利用边界跟踪算法获取对象边界。在对象边界图像上,局部应用Hough变换检测大腿和小腿的直线,从而得到大腿和小腿的倾斜角。用最小二乘法将一个周期内的倾斜角序列拟合成5阶多项式,把Fourier级数展开后得到的相位与振幅的乘积定义为低维步态特征向量。在小样本的数据库上用F isher线性分类器验证所研究算法的性能,正确分类率为79.17%。在步态数据库不很理想的情况下也获得了较好的识别率。

Gait is an emergent biometric aimed essentially to recognize people by the way they walk.Gait as a biometric can be seen as advantageous over other forms of biometric identification techniques,for it offers the possibility to identify people at a distance without any interaction or co-operation from the subject.This paper proposes a novel automatic gait recognition method,which extracts gait signature from legs of the subject.For each image sequence,background subtraction based on chromaticity distortion is used to segment moving objects.Boundary tracking algorithm is then used to find perimeter pixels in each processed binary image sequence.This paper makes use of Hough Transform to locally extract the lines which represent legs,and thus obtains inclination angles of upper legs and lower legs.The angles are then fitted to a fifth-order polynomial by least squares method.The polynomial curve is expressed by a Fourier series.The lower-dimensional gait signature vector,that is,the product of phase and magnitude,is derived from phase and magnitude spectra.Fisher Linear Classifier is used to validate the performance of the proposed algorithm on small database samples and the correct classification rate is 79.17%.The recognition rate is still good for these unideal outdoor image sequences.