基于内容的图像检索中，高维特征向量的有效降维是一项关键技术，流形学习算法LLE 是有效的非线性降维方法，而SVM 的泛化能力较强，本文将二者结合在一起，成为一种新的有监督流形学习算法SLLESVM，并将其应用在图像数据的检索上，效果良好。关键词: 有监督; 流行学习; SLLESVM; 图像检索Abstract: In Content-Based Image Retrieval, the dimension reduction of high dimensioneigenvector is a key technology, manifold learning algorithm LLE is an effective non-linear rdimension reduction method, and SVM has stronger generalization ability, this paper combined two methods, became a new supervised manifold learning algorithm SLLESVM, and applied it in the retrieval of image data, the experiments proved that the results are good.Key words: supervised; manifold learning; SLLESVM; image retrieval