A Futuristic Hybrid Image Retrieval System based on an Effective Indexing Approach for Swift Image Retrieval

Abstract

As the multimedia content is increasing rapidly, there is an urgent need of an effective image retrieval system and the key to this peculiar problem is denoted by Content-based image retrieval (CBIR) system. But, for retrieving some particular images from a massive database, the retrieval process becomes time consuming. So, in order to reduce the retrieval time, image indexing is utilized. The present work highlights an effective image retrieval system with an indexing technique to reduce the retrieval time. This paper focuses on the formation of a hybrid image retrieval system in which texture, color and shape attributes of an image are withdrawn by using gray level cooccurrence matrix (GLCM), color moment and region props procedure respectively. Then, extracted fused features are optimally selected by using principal component analysis (PCA). Afterwards, two types of Indexing techniques namely, similaritybased indexing and cluster-based indexing have been tested on the developed hybrid system to find the best amongst them. The results of the hybrid color descriptor based on Cluster-based Indexing technique depict that the proposed system has enhanced results. Average precision of 93.8%, 79.6%, 70%, 98.7%, 93.5% and 79.5% has been obtained on Corel-1K, Corel5K, Corel-10K, COIL-100, GHIM-10 and ZUBUD datasets respectively.