基于改进型深度数据流形的数据分类算法及在人脸中的应用

A Novel Data Classification Algorithm and Application Research Based on Modified Depth Data Manifold

针对数据分类问题的局限,提出一种基于改进型深度数据流形的数据分类算法并将其应用到人脸识别中。首先,通过采集人脸图像的深度信息,利用稀疏表示对其进行去噪处理;再结合图像的颜色信息,重新生成三维人脸信息数据库,通过对人脸数据的流形分析得到最优的降维结果,按十字十乘交叉验证法的原则选取训练集和测试集,将训练集输入支持向量机算法建立数据分类器;最后,将测试集输入训练完成的分类器中,实现人脸数据分类。选取ORL、Yale两类人脸图像标准数据库与传统人脸识别算法进行交叉对比实验,验证算法的优越性和可行性。实验结果表明:所提出的算法有较高的分类准确率,可有效地完成人脸识别。

For the localization of data classification, a novel data classification algorithm based on modified data manifold is proposed. It is used as the method of face recognition. Firstly, the depth information of images are collected by Kinect, and the sparse representation can be used to do the denoising. Secondly, the three-dimensional face data base can be established by the colour information and depth information. The dimension of data sets is reduced by the analysis of the data manifold, and optimal results of data dimension reduction can be gotten. The training and test sets are gotten by the principle of ten cross validation, and data classifier can be gotten by the support vector machine. Finally, the test sets are inputted, and the face data classification can be achieved. The two classes of data sets are selected as the experimental data, which consist of ORL and Yale. The comparison experiments can be achieved by the two data sets, and the experiment results show that the proposed method not only has a higher classification accuracy rate,but it has a great effect to achieve face recognition.