基于监督的局部保存投影的人脸识别

Supervised Locality Preserving Projection Based Face Recognition

人脸识别是计算机视觉的一个重要分支,而且它可以被应用到许多领域。维数约减对人脸识别来说是十分重要的,不仅可以去除一些不必要的噪声,还可以有效地减少计算复杂度。传统的局部保存投影(LPP)算法仅仅考虑了样本点的局部邻居信息,并没有考虑到不同类别的样本对分类效果的影响,结果会造成不同类数据点的重叠。为减少计算的复杂度,在局部保存投影算法的基础上引入样本的类别信息,构造了监督的局部保存投影算法(SLPP)。通过人脸识别仿真实验,验证了新算法的有效性与健壮性,并改变上述的不足。

Face recognition is an important branch of computer vision and can be applied in lots of fields.In the past few years,face recognition has received more focuses.Dimensionality reduction is the key technique for face recognition.It not only can clear some unnecessary noise,also can reduce the complicated degree of computation effectively.The traditional Locality Preserving Projection algorithm only considers the local information of the example data and neglects the impact of the different class labels,which results in the overlap of the data points of different classes.In this paper,we introduce the class labels information based on LPP,and propose Supervised Locality Preserving Projections（SLPP）.Through some face recognition experiments,we show the effectiveness and robustness of SLPP.