Efficient Image Retrieval System based on Differential Latent Semantic Index

S. Nilufar, L. Chen (Canada), and A.R. Shoyeb Ahmed Siddique (Bangladesh)


Latent semantic index, perceptual relation, image clusters, likelihood function.


This paper presents and evaluates differential latent semantic indexing method for content based image retrieval by constructing perceptual knowledge from collections of images. It also proposes a novel technique for uncovering perceptual relationships among concepts on the basis of image feature similarity and statistics between image classes. DLSI scheme which depicts much richer knowledge, capturing more unique characteristics of particular image, builds relationship between the perceptual concepts in the form of intra and extra differential feature space. Experimental result gives the supporting proof of the effectiveness of DLSI approach for image retrieval.

Important Links:

Go Back