Learning contextual dissimilarity on tensor product graph for visual re-ranking

作者：

Highlights：

• We introduce the mean first-passage time (MFPT) as the contextual dissimilarity.

• The cost function of MFPT is generated with a hybrid fitting constraint.

• We propose a new contextual dissimilarity based on MFPT and tensor product graph.

• The re-ranking result of our methods outperforms other algorithms.

摘要

•We introduce the mean first-passage time (MFPT) as the contextual dissimilarity.•The cost function of MFPT is generated with a hybrid fitting constraint.•We propose a new contextual dissimilarity based on MFPT and tensor product graph.•The re-ranking result of our methods outperforms other algorithms.

论文关键词：Diffusion process,Contextual dissimilarity,Tensor product graph,Mean first-passage time,Hybrid fitting constraint

论文评审过程：Received 25 April 2017, Revised 1 June 2018, Accepted 21 June 2018, Available online 4 July 2018, Version of Record 5 September 2018.

论文官网地址：https://doi.org/10.1016/j.imavis.2018.06.006