Saliencies and Symmetries: Toward 3D Object Recognition from Large Model Databases
Abstract
The construction of interpretation tables from database models is introduced, and a recognition procedure using scene feature groups is discussed. Techniques for extraction of feature group equivalence classes and computation of feature group saliency are discussed. Two methods to reduce the computational burdens associated with a large model database are proposed and tested on polyhedral objects. The first method reduces the population of protohypotheses in the interpretation tables consulted during recognition by excluding redundant feature groups produced from object symmetries. The second method assigns a population-based numerical measure of saliency to each feature group retrieved from the scene; this measure allows only the most salient feature groups to be used in object recognition.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">></ETX>
Cite
Text
Flynn. "Saliencies and Symmetries: Toward 3D Object Recognition from Large Model Databases." IEEE/CVF Conference on Computer Vision and Pattern Recognition, 1992. doi:10.1109/CVPR.1992.223256Markdown
[Flynn. "Saliencies and Symmetries: Toward 3D Object Recognition from Large Model Databases." IEEE/CVF Conference on Computer Vision and Pattern Recognition, 1992.](https://mlanthology.org/cvpr/1992/flynn1992cvpr-saliencies/) doi:10.1109/CVPR.1992.223256BibTeX
@inproceedings{flynn1992cvpr-saliencies,
title = {{Saliencies and Symmetries: Toward 3D Object Recognition from Large Model Databases}},
author = {Flynn, Patrick J.},
booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition},
year = {1992},
pages = {322-327},
doi = {10.1109/CVPR.1992.223256},
url = {https://mlanthology.org/cvpr/1992/flynn1992cvpr-saliencies/}
}