Using Latent Semantic Index for Content-Based Image Retrieval

Andy, Bernardus Ari Kuncoro

Abstract: In this paper, the latent semantic indexing (LSI) based method was used to the various image feature extracted matrix in order to perform content-based image retrieval. The feature extraction techniques include color histogram, color auto-gram, color moment, gray-scale, and wavelet moment. The implementation of LSI here is to achieve an improved image retrieval performance, because it reduces the size of input image matrix by determining parameter of k. The objective of this paper is to know the performance of LSI-based CBIR using precision and recall parameters for each feature extractions.

Keywords: image feature extractions, latent semantic indexing, content-based image retrieval