Generalized Dynamic Programming Approaches for Object Detection: Detecting Spine Boundaries and Vertebra Endplates

Abstract

Object detection employing high-level knowledge is a challenging problem in image analysis. The authors propose a dynamic programming approach to address some related issues in this aspect. In particular, we propose to fuse the detection of two curves to form a dual dynamic programming procedure so that spatial relationships between the two curves can be enforced. In another pursuit of applying object level knowledge, we propose to introduce local backward tracing to the forward propagation step in dynamic programming, so that global constraints, containing even unknown parameters, can be imposed in a progressive manner. These approaches are explained in the context of spine landmark detection. Experimental results are presented to show the efficiency of the proposed methods. The methods are extendable to other application domains.

Cite

Text

Wei et al. "Generalized Dynamic Programming Approaches for Object Detection: Detecting Spine Boundaries and Vertebra Endplates." IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2001. doi:10.1109/CVPR.2001.990632

Markdown

[Wei et al. "Generalized Dynamic Programming Approaches for Object Detection: Detecting Spine Boundaries and Vertebra Endplates." IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2001.](https://mlanthology.org/cvpr/2001/wei2001cvpr-generalized/) doi:10.1109/CVPR.2001.990632

BibTeX

@inproceedings{wei2001cvpr-generalized,
  title     = {{Generalized Dynamic Programming Approaches for Object Detection: Detecting Spine Boundaries and Vertebra Endplates}},
  author    = {Wei, Guo-Qing and Qian, Jian Zhong and Schramm, Helmuth},
  booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  year      = {2001},
  pages     = {I:954-959},
  doi       = {10.1109/CVPR.2001.990632},
  url       = {https://mlanthology.org/cvpr/2001/wei2001cvpr-generalized/}
}