TY - GEN
T1 - Efficient geometric re-ranking for mobile visual search
AU - Luo, Junwu
AU - Lang, Bo
PY - 2013
Y1 - 2013
N2 - The state-of-the-art mobile visual search approaches are based on the bag-of-visual-word (BoW). As BoW representation ignores geometric relationship among the local features, a full geometric constraint like RANSAC is usually used as a post-processing step to re-rank the matched images, which has been shown to greatly improve the precision but at high computational cost. In this paper we present a novel and efficient geometric re-ranking method. Our basic idea is that the true matching local features should be not only in a similar spatial context, but also have a consistent spatial relationship, thus we simultaneously introduce context similarity and spatial similarity to describe the geometric consistency. By incorporating these two geometric constraints, the co-occurring visual words in the same spatial context can be regarded as a "visual phrase"and significantly improve the discriminative power than single visual word. To evaluate our approach, we perform experiments on Star5k and ImageNet100k dataset. The comparison with the BoW method and Soft-assignment method highlights the effectiveness of our approach in both accuracy and speed.
AB - The state-of-the-art mobile visual search approaches are based on the bag-of-visual-word (BoW). As BoW representation ignores geometric relationship among the local features, a full geometric constraint like RANSAC is usually used as a post-processing step to re-rank the matched images, which has been shown to greatly improve the precision but at high computational cost. In this paper we present a novel and efficient geometric re-ranking method. Our basic idea is that the true matching local features should be not only in a similar spatial context, but also have a consistent spatial relationship, thus we simultaneously introduce context similarity and spatial similarity to describe the geometric consistency. By incorporating these two geometric constraints, the co-occurring visual words in the same spatial context can be regarded as a "visual phrase"and significantly improve the discriminative power than single visual word. To evaluate our approach, we perform experiments on Star5k and ImageNet100k dataset. The comparison with the BoW method and Soft-assignment method highlights the effectiveness of our approach in both accuracy and speed.
UR - https://www.scopus.com/pages/publications/84875970722
U2 - 10.1007/978-3-642-37484-5_42
DO - 10.1007/978-3-642-37484-5_42
M3 - 会议稿件
AN - SCOPUS:84875970722
SN - 9783642374838
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 520
EP - 532
BT - Computer Vision - ACCV 2012 International Workshops, Revised Selected Papers
PB - Springer Verlag
T2 - 2012 International Workshop on Computer Vision with Local Binary Pattern Variants, LBP 2012, 2012 Workshop on Computational Photography and Low-Level Vision, 2012 Workshop on Developer-Centered Computer Vision, 2012 Workshop on Background Models Challenge, BMC 2012, 2012 Workshop on e-Heritage, 2012 Workshop on Color Depth Fusion in Computer Vision, 2012 Workshop on Face Analysis: The Intersection of Computer Vision and Human Perception, 2012 Workshop on Detection and Tracking in Challenging Environments, DTCE 2012, 2012 International Workshop on Intelligent Mobile Vision, IMV 2012, held in conjunction with the 11th Asian Conference on Computer Vision, ACCV 2012
Y2 - 5 November 2012 through 6 November 2012
ER -