基于支持向量机和特征向量提取的人脸识别框架

为了研究支持向量机在人脸识别中的应用,提出了人脸识别框架,该框架首先利用特征向量提取算法对人脸进行特征提取,得到人脸面部纹理特征数据,然后采用支持向量机对提取的数据进行训练,得到人脸模板,并依据人脸模板对人脸进行识别。实验采用ORL人脸数据库作为实验数据,使用LBP算法提取特征向量,使用LIBSVM训练得到人脸模板数据库,当模板人脸数与预测人脸数比值一定时,随着人数增加,其预测的正确率会有所下滑;当人数一定时,人脸模板与预测人脸数值上升,其正确率会有所上升。当选择一个相对合适的模板比例时,正确率将会达到89.29%以上。实验结果表明,提出的框架对于人脸具有良好的识别能力。

Face recognition framework based on support vector machine and feature vector extraction

In order to study the application of support vector machine in face recognition, a face recognition framework is proposed. This framework firstly extracts face feature with the feature vector extraction algorithm to get the face feature data, and trains the data to obtain a face model, then recognizes the faces according to the face model. The experiment uses ORL face database as experimental data, uses LBP to extract the face feature, and uses LIBSVM to train the data to get the face model. When the ratio of the number of faces in templates to the prediction number of faces is the same, the prediction accuracy will decrease with the increase of the number; when the number of faces is the same, the prediction accuracy will increase with the increase of the face templates and the prediction number of faces. When the proportion of the template is relatively appropriate, the accuracy will exceed 89.29%. The experimental results show that this framework works well in recognizing faces.