Encoding multiple contextual clues for partial-duplicate image retrieval

Zhili Zhou, Q. M. Jonathan Wu, Xingming Sun

Abstract: Highlights•The existing approaches ignore the visual relationships and spatial configurations between visual words.•Encoding multiple contextual clues can improve the discriminability of BOW representation.•The proposed asymmetrical context selection strategy is favorable for retrieval accuracy.•Our approach provides better retrieval accuracy than the state-of-the-arts.•It also achieves comparable time and space efficiency to the state-of-the-arts.