Part aggregation in a compositional model based on the evaluation of feature cooccurrence statistics (bibtex)
by Martin Stommel, Klaus-Dieter Kuhnert
Abstract:
In this paper an appearance based, compositional approach to the recognition of deformable objects is presented. First, a hierarchical object model is proposed. On di erent levels of abstraction the model represents object categories, di erent views of an object, the parts of an object and basic feature vectors. Then, a training method based on multiple clustering steps is described. This paper addresses in particular the aggregation of features to parts and provides a statistical justi cation for feature clustering on the lowest level of the hierarchy. The performance of the proposed methods is demonstrated on a cartoon data base, where a high accuracy of 80\% is achieved.
Reference:
Martin Stommel, Klaus-Dieter Kuhnert, "Part aggregation in a compositional model based on the evaluation of feature cooccurrence statistics", In Int?l Conf. on Image and Vision Computing New Zealand, IEEE, Christchurch, New Zealand, 2008.
Bibtex Entry:
@INPROCEEDINGS{Stommel2008b,
  author = {Stommel, Martin and Kuhnert, Klaus-Dieter},
  title = {Part aggregation in a compositional model based on the evaluation
	of feature cooccurrence statistics},
  booktitle = {Int?l Conf. on Image and Vision Computing New Zealand},
  year = {2008},
  editor = {Irie, Kenji and Pairman, David},
  address = {Christchurch, New Zealand},
  month = {November26--28},
  publisher = {IEEE},
  abstract = {In this paper an appearance based, compositional approach to the recognition
	of deformable objects is presented. First, a hierarchical object
	model is proposed. On dierent levels of abstraction the model represents
	object categories, dierent views of an object, the parts of an object
	and basic feature vectors. Then, a training method based on multiple
	clustering steps is described. This paper addresses in particular
	the aggregation of features to parts and provides a statistical justication
	for feature clustering on the lowest level of the hierarchy. The
	performance of the proposed methods is demonstrated on a cartoon
	data base, where a high accuracy of 80{\%} is achieved.},
  doi = {10.1109/IVCNZ.2008.4762081},
  isbn = {978-1-4244-2582-2},
  owner = {pmania},
  timestamp = {2012.11.06},
  url = {http://ieeexplore.ieee.org/stamp/stamp.jsp?tp={\&}arnumber=4762081}
}
Powered by bibtexbrowser