Visual Reranking with Improved Image Graph

TL;DR

This paper presents an improved image reranking method using a robust directed image graph and feature fusion, significantly enhancing image search accuracy on benchmark datasets.

Contribution

The paper introduces a novel directed image graph for reranking that is robust to outliers and incorporates rank-level feature fusion, including color information, to improve image search results.

Findings

Significant performance improvements on benchmark datasets

Effective integration of multiple features for reranking

Competitive results compared to state-of-the-art methods

Abstract

This paper introduces an improved reranking method for the Bag-of-Words (BoW) based image search. Built on [1], a directed image graph robust to outlier distraction is proposed. In our approach, the relevance among images is encoded in the image graph, based on which the initial rank list is refined. Moreover, we show that the rank-level feature fusion can be adopted in this reranking method as well. Taking advantage of the complementary nature of various features, the reranking performance is further enhanced. Particularly, we exploit the reranking method combining the BoW and color information. Experiments on two benchmark datasets demonstrate that ourmethod yields significant improvements and the reranking results are competitive to the state-of-the-art methods.