The Community for Technology Leaders
2013 IEEE Conference on Computer Vision and Pattern Recognition (2012)
Providence, RI USA
June 16, 2012 to June 21, 2012
ISSN: 1063-6919
ISBN: 978-1-4673-1226-4
pp: 726-732
S. Escalera , Comput. Vision Center, Barcelona, Spain
P. Radeva , Comput. Vision Center, Barcelona, Spain
D. Dimov , Inst. of Inf. & Commun. Technol., Sofia, Bulgaria
M. Reyes , Comput. Vision Center, Barcelona, Spain
A. Marinov , Inst. of Inf. & Commun. Technol., Sofia, Bulgaria
A. Hernandez-Vela , Comput. Vision Center, Barcelona, Spain
N. Zlateva , Inst. of Inf. & Commun. Technol., Sofia, Bulgaria
ABSTRACT
We present a generic framework for object segmentation using depth maps based on Random Forest and Graph-cuts theory, and apply it to the segmentation of human limbs in depth maps. First, from a set of random depth features, Random Forest is used to infer a set of label probabilities for each data sample. This vector of probabilities is used as unary term in α-β swap Graph-cuts algorithm. Moreover, depth of spatio-temporal neighboring data points are used as boundary potentials. Results on a new multi-label human depth data set show high performance in terms of segmentation overlapping of the novel methodology compared to classical approaches.
INDEX TERMS
probability, graph theory, image motion analysis, image segmentation, optimisation, spatio-temporal neighboring data points, graph cuts optimization, multilimb human segmentation, depth maps, object segmentation, random forest, random depth features, label probabilities, α-β swap graph-cuts algorithm, Humans, Radio frequency, Image segmentation, Vegetation, Training, Vectors, Joints
CITATION
S. Escalera, P. Radeva, D. Dimov, M. Reyes, A. Marinov, A. Hernandez-Vela, N. Zlateva, "Graph cuts optimization for multi-limb human segmentation in depth maps", 2013 IEEE Conference on Computer Vision and Pattern Recognition, vol. 00, no. , pp. 726-732, 2012, doi:10.1109/CVPR.2012.6247742
168 ms
(Ver 3.3 (11022016))