Image Retrieval Using Locality Preserving Projections: A Recent Study

Abstract

Image retrieval algorithms are important in web surfing. There are many algorithms available, but they all have disadvantages and low retrieval rates. This work proposes a novel method for image retrieval called locality preserving projections (LPPs).The benefits of the LPP approach are that LPP is a linear technique that retains the image space’s local info. The method has a higher retrieval rate since it uses principal component analysis, nearest neighbour, and eigen features to extract hyper parameters for classification. This unique approach is tested using the Yale facial image dataset, and the algorithm's performance is evaluated using several classifiers such as Decision Tree.