Enhancing Web Image Retrieval Precision: A Hybrid Approach with Click-Driven Re-Ranking

Abstract

In image search re-ranking, a major problem restricting image retrieval development is an intent gap, which is a gap between the user’s real intent and query/demand representation, besides the well-known semantic gap. In the past, for achieving effective web image retrieval, classifier space, or feature space, was explored by researchers. Visual information and image initial ranks with a single feature are only considered in conventional re-ranking techniques for measuring typicality and similarity in web image retrieval, while overlooking click-through data influence. For image retrieval, various image feature aggregation methods have shown their effectiveness in recent days. But uplifting the best features impact for a specific query image presents a major challenge in computer vision problems. In this paper, based on web queries, features are assigned weights, where different weights are received by different queries in a ranked list. The IABC algorithm used to compute weights is a data-driven algorithm that does not require any learning. Finally, in a web, color and texture features are fused using fusion, and these features are extracted with respective modalities. A Hypergraph Construction Clustering (HCC) re-ranking with click-based similarity and typicality procedure termed HCCCST is used in the re-ranking technique. Its operation depends on the selection of click-based triplets, and a classifier is used for integrating multiple features into a unified similarity space. The web image search re-ranking performance is greatly enhanced using the proposed technique.