基于支持向量机的步态识别算法研究

为了准确快速地进行运动人体的步态识别,提出了一种基于主分量分析(PCA)和统一Hu矩融合的步态识别算法。将人体髋关节以下作为感兴趣区域,对图像序列中运动人体的感兴趣区域进行了分割,并提取主分量外形特征,同时计算感兴趣区域的统一Hu不变矩特征,将二者结合,构成步态序列的特征空间,采用支持向量机(SVM)分类器进行分类识别,通过MATLAB仿真实验验证了算法的有效性。实验结果表明,该算法识别速度快,具有较高的识别率。

Gait Recognition Based on Support Vector Machine

In order to increase the accuracy and rate of human gait recognition,an algorithm of gait recognition is proposed,which is based on a fusion of principal component analysis(PCA) and unified Hu moments,and takes the following part of hip joint as the interested area.First,the interested area of body in each image sequence is segmented,then the principal component analysis appearance features and unified Hu invariant moments of the interested area are extracted,which are then combined together into one feature space,and support vector machine(SVM) is used for classification,then the effectiveness of the algorithm is verified by MATLAB simulation experiments.The result shows that the algorithm achieves higher recognition speed and recognition rate.