Abstract:

This paper proposes a novel and simple method of gait feature extraction. Body silhouette extraction is achieved by background subtraction. The body silhouette is projected along its midline to obtain the fore-and-aft midline projection vectors, which are then combined into one dimensional data vector as the gait feature. PCA is applied to reduce data dimension, and gait classification and recognition is performed by support vector machine. The result of the experiment demonstrates that the approach is effective and has encouraging recognition performance.

Key words: Feature extraction, Midline projection, Principal component analysis, Support vector machine

摘要：

提出了一种新颖且简单有效的步态特征提取方法。用背景差方法得到运动人体的轮廓，对提取出来的轮廓沿中线进行投影，将二维的数据压缩为一维的中线向量。用PCA和SVM相结合的方法对其进行了验证，结果证明了该特征提取方法的有效性及良好的识别效果。

关键词: 特征提取, 中线投影, 主成分分析, 支持向量机