Research on the Optimization of Locality Preserving Projections

Locality Preserving Projections (LPP) is based on a nearest neighbor graph which models the local geometrical structure of the image manifold. However, this graph can not always accurately estimate the intrinsic manifold structure. A novel algorithm called Iterative locality preserving projections (ILPP) is preposed. ILPP iteratively updates the nearest neighbor graph, so that it can better model the intrinsic manifold structure. Experimental results comparison show that our algorithm outperforms PCA and LPP for image retrieval.