半监督局部线性嵌入算法在人脸识别中的应用

提高人脸识别算法的识别率,提出一种基于半监督局部线性嵌入(Semi-Supervised Locally Linear Embedding,SSLLE)的人脸图像识别方法。针对局部线性嵌入(Locally Linear Embedding,LLE)算法非监督学习的缺陷,引入半监督思想,在构造邻域的时候利用部分样本的标签信息来重新调整距离矩阵;使用调整后的距离矩阵进行线性重建从而实现数据降维。在Yale和ORL人脸库上的实验结果表明,能有效的提高人脸识别的性能。

Application of Semi-supervised Locally Linear Embedding in Face Recognition

In order to improve the recognition rate of face recognition algorithm,a novel face recognition method based on Semi-Supervised Locally Linear Embedding is proposed. For the defects of the unsupervised learning of Locally Linear Embedding algorithm,a semi-supervised thought is introduced. When constructing neighborhood,using some labeled samples to readjust the distance matrix; then reconstruct the samples linearly by using the adjusted matrix in order to achieve data dimensionality reduction. The experimental results on Yale and ORL face database show that the proposed method can effectively improve the performance of face recognition.