Tree-walk kernels for computer vision - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Book Sections Year : 2012

Tree-walk kernels for computer vision


We propose a family of positive-definite kernels between images, allowing to compute image similarity measures respectively in terms of color and of shape. The kernels consists in matching subtree-patterns called "tree-walks" of graphs extracted from the images, e.g. the segmentation graphs for color similarity and graphs of the discretized shapes or the point clouds in general for shape similarity. In both cases, we are able to design computationally efficient kernels which can be computed in polynomial-time in the size of the graphs, by leveraging specific properties of the graphs at hand such as planarity for adjacency graphs (segmentation graphs) or factorizability of the associated graphical model for point clouds. Our kernels can be used by any kernel-based learning method, and hence we present experimental results for supervised and semi-supervised classification as well as clustering of natural images and supervised classification of handwritten digits and Chinese characters from few training examples.
Fichier principal
Vignette du fichier
hb_2012_treewalks_cv_chapter_hal.pdf (312.51 Ko) Télécharger le fichier
Vignette du fichier
treewalks.png (30.84 Ko) Télécharger le fichier
Origin : Explicit agreement for this submission
Format : Figure, Image

Dates and versions

hal-00756815 , version 1 (23-11-2012)


  • HAL Id : hal-00756815 , version 1


Zaid Harchaoui, Francis Bach. Tree-walk kernels for computer vision. Lezoray, Olivier and Grady, Leo. Image Processing and Analysis with Graphs: Theory and Practice, CRC Press, pp.499-525, 2012, Digital Imaging and Computer Vision Series, 978-1-439-85507-2. ⟨hal-00756815⟩
482 View
754 Download


Gmail Facebook X LinkedIn More