Multimedia plays an important role in the web-based learning environment. In this paper, we validate management and retrieval of large multimedia collections through high-level semantics. A novel algorithm is proposed for automatic annotation of image based on support vector machine and statistical learning. In addition, we construct cross-media indexing for multi-modal data upon the annotation result to support cross-media search. Experiments show that our algorithm can interpret multimedia semantics accurately and cross-media indexing can support cross-media search effectively.