基于SVM和KNR的体型分类算法

Body Shapes Classification Algorithm Based on SVM and KNR

本文提出一种基于人体宽高比的体型分类算法。首先提取人体主躯干(肩宽、体长)数据,减小衣着、手臂对体型的影响,然后以主躯干的宽高比为特征,尝试用KNR(核非线性回归)实现体型分类,并与SVM(支持向量机)的性能相比较。实验结果表明,KNR总体取得了较好的分类效果。

This paper proposes a body type classification algorithm based on the aspect ratio of human body.Firstly,the main body(shoulder width,body length) data were extracted to reduce the influence of clothing and arms on body type.Then,the body type was classified by KNR(nuclear nonlinear regression) and compared with SVM(support vector machine) based on the aspect ratio of main body.The experimental results show that the KNR has achieved a good classification effect.