Deformable Object Tracking Based on Max-pooling Graph Matching
WANG Zhi-dan;JIANG Jian-guo;QI Mei-bin;School of Computer and Information,Hefei University of Technology;
This paper develops a novel deformable object tracking algorithm based on max-pooling graph matching,which can be applied in the scenes with large deform ations and severe occlusions. The dynam ic graph is built based on candidate parts extracted by over-segm entation m ethod from searching area,nam ely feature representation of candidate parts and geom etric structure between them. Based on m ax-pooling graph m atching m ethod,the m atching relations between target parts and candidate parts are found to calculate the confidence m ap of target location. Considering both the support of holistic target and local parts,the optim al target location can be determ ined. Com pared to state-of-the-art m ethods,experim ental results on several deform able sequences dem onstrate the effectiveness and robustness of the proposed m ethod.
CAJViewer7.0 supports all the CNKI file formats; AdobeReader only supports the PDF format.