VLAD re-ranking: Iteratively estimating the probability of relevance with relationships between dataset images

作者：

Highlights：

• Our method is an efficient framework which combines BOW and VLAD for better accuracy.

• We regard the problem of similarity evaluation as probability evaluation.

• Probability estimation is formulated with the application of the image graph.

• Our iterative evaluation procedure further improves the retrieval accuracy.

摘要

Highlights•Our method is an efficient framework which combines BOW and VLAD for better accuracy.•We regard the problem of similarity evaluation as probability evaluation.•Probability estimation is formulated with the application of the image graph.•Our iterative evaluation procedure further improves the retrieval accuracy.

论文关键词：Object retrieval,Image graph,Re-ranking,Bayes theory

论文评审过程：Received 27 September 2015, Revised 10 March 2016, Accepted 10 March 2016, Available online 17 March 2016, Version of Record 23 March 2016.

论文官网地址：https://doi.org/10.1016/j.image.2016.03.004