Locating Objects of Varying Shape Using Statistical Feature Detectors
Abstract
Most deformable models use a local optimisation scheme to locate their targets in images, and require a 'good enough' starting point. This paper describes an approach for generating such starting points automatically given no prior knowledge of the pose of the target(s) in the image. It relies upon choosing a suitable set of features, candidates for which can be found in the image. Hypotheses are formed from sets of candidates, and their plausibility tested using the statistics of their relative positions and orientations. The most plausible are used as the initial position of an Active Shape Model, which can then accurately locate the target object. The approach is demonstrated for two different image interpretation problems.
Cite
Text
Cootes and Taylor. "Locating Objects of Varying Shape Using Statistical Feature Detectors." European Conference on Computer Vision, 1996. doi:10.1007/3-540-61123-1_161Markdown
[Cootes and Taylor. "Locating Objects of Varying Shape Using Statistical Feature Detectors." European Conference on Computer Vision, 1996.](https://mlanthology.org/eccv/1996/cootes1996eccv-locating/) doi:10.1007/3-540-61123-1_161BibTeX
@inproceedings{cootes1996eccv-locating,
title = {{Locating Objects of Varying Shape Using Statistical Feature Detectors}},
author = {Cootes, Timothy F. and Taylor, Christopher J.},
booktitle = {European Conference on Computer Vision},
year = {1996},
pages = {465-474},
doi = {10.1007/3-540-61123-1_161},
url = {https://mlanthology.org/eccv/1996/cootes1996eccv-locating/}
}