Image retrieval with graph kernel on regions

In the framework of the interactive search in image databases, we are interested in similarity measures able to learn during the search and usable in real-time. Images are represented by adjacency graphs of regions. In order to compare attributed graphs, we employ kernels on graphs built on sets of paths. In this paper, we introduce a fast kernel function whose similarity is based on several matches. We also introduce new features for edges in the graph. Experiments on a specific database having objects with heterogeneous backgrounds show the performance of our object retrieval technique.

[1]  F. Suard,et al.  Pedestrian detection using stereo-vision and graph kernels , 2005, IEEE Proceedings. Intelligent Vehicles Symposium, 2005..

[2]  Sylvie Philipp-Foliguet,et al.  FReBIR: An image retrieval system based on fuzzy region matching , 2009, Comput. Vis. Image Underst..

[3]  Sébastien Sorlin,et al.  Mesurer la similarité de graphes , 2006 .

[4]  Trevor Darrell,et al.  The pyramid match kernel: discriminative classification with sets of image features , 2005, Tenth IEEE International Conference on Computer Vision (ICCV'05) Volume 1.

[5]  Frédéric Jurie,et al.  Creating efficient codebooks for visual recognition , 2005, Tenth IEEE International Conference on Computer Vision (ICCV'05) Volume 1.

[6]  Hisashi Kashima,et al.  Kernel-based discriminative learning algorithms for labeling sequences, trees, and graphs , 2004, ICML '04.