摘要
步态识别是一种基于步态特征的生物特征识别技术,包括身高和体形等。步态特征中的傅里叶描述子具有尺度、平移和旋转不变性,可以表示人体的轮廓特征。本文提出了一种识别方法:首先运用粒子滤波算法得到更准确的目标位置信息,在此基础上计算傅里叶描述子,进而实现更优越的步态识别效果。通过对滤波前后的傅里叶特征与真实目标特征的对比,证明此方法提取的特征信息更为准确,从而提高对人体目标的识别效率。
Gait recognition is a new technology of recognition based on gait character which belongs to the biologic character and includes height,shape and so on.Fourier descriptor has the advantages that the scale,translation and rotation are invariance,so it can be expressed as the outline characteristics of the human body.A method is described in this paper：after tracking the target and getting more accurate position information in the video sequence with particle filter,extract the gait features of the target-Fourier descriptors,and then make better identification.Comparing the characteristics of the real goal with the ones got before or after the Fourier filtering,it proves that this method extracts more accurate feature information,thus recognition rate will be increased correspondingly.
出处
《电气电子教学学报》
2010年第3期35-38,共4页
Journal of Electrical and Electronic Education
关键词
粒子滤波
傅里叶描述子
特征提取
particle filter
Fourier descriptors
feature extraction
