Image Retrieval Based on PCA-LPP

In many areas of image retrieval, pattern recognition and data mining, one is ofen confronted with some form of dimensionality reduction. In this paper, we introduce a algorithm of Principal Components Analysis-Locality Preserving Projections ( PCA-LPP ). This algorithm can compute the principal components of a sequence of data set and at the time optimally preserves the neighborhood structure of the data set. This procedure is done by merging the runs of two algorithms based on principal component analysis (PCA) and Locality Preserving projections (LPP) running sequentially. This algorithm is applied to image retrieval problem. Simulation results on Corel image database showed better accuracy rate of this algorithm compared to PCA and LPP algorithms.