作　　者： ;

机构地区： 广东工业大学自动化学院

出　　处： 《计算机工程与设计》 2012年第4期1542-1546,共5页

摘　　要： 为进一步提高红外步态识别精度,构建了一种多分类器融合识别新模型,在根据各单分类器识别输出值构建度量向量的基础上,进行基于粗糙集支持向量机的多分类器融合识别。通过在Matlab7.5平台利用中科院红外步态库进行识别仿真实验,获得识别率和累积匹配分值的实验数据及对比结果。实验结果表明,基于粗糙集支持向量机的多分类器融合识别模型比单分类器在识别率方面有大幅度提高,识别性能理想,识别精度高。 A new multiple classifier fusion recognition model is constructed for further improving the accuracy of the infrared gait recognition. Multi-classifier fusion recognition is implemented by using the rough set support vector machine on the basis of measurement vector constructed by the output value of single classifier recognition. The recognition emulation test is executed by using the infrared gait database of the Chinese Academy of Sciences on the MatlabT. 5 platform. The laboratory data of the recog- nition rate and the cumulative match score are obtained. The conclusion is acquired that the multiple classifier fusion recognition is higher precision, better recognition performance and bigger recognition rate than the single classifier recognition.

关 键 词： 粗糙集 支持向量机 红外 步态识别 多分类器融合

领　　域： [自动化与计算机技术] [自动化与计算机技术]