Multiresolution Tangent Distance for Affine-Invariant Classification
Abstract
The ability to rely on similarity metrics invariant to image transforma(cid:173) tions is an important issue for image classification tasks such as face or character recognition. We analyze an invariant metric that has performed well for the latter - the tangent distance - and study its limitations when applied to regular images, showing that the most significant among these (convergence to local minima) can be drastically reduced by computing the distance in a multiresolution setting. This leads to the multi resolution tangent distance, which exhibits significantly higher invariance to im(cid:173) age transformations, and can be easily combined with robust estimation procedures.
Cite
Text
Vasconcelos and Lippman. "Multiresolution Tangent Distance for Affine-Invariant Classification." Neural Information Processing Systems, 1997.Markdown
[Vasconcelos and Lippman. "Multiresolution Tangent Distance for Affine-Invariant Classification." Neural Information Processing Systems, 1997.](https://mlanthology.org/neurips/1997/vasconcelos1997neurips-multiresolution/)BibTeX
@inproceedings{vasconcelos1997neurips-multiresolution,
title = {{Multiresolution Tangent Distance for Affine-Invariant Classification}},
author = {Vasconcelos, Nuno and Lippman, Andrew},
booktitle = {Neural Information Processing Systems},
year = {1997},
pages = {843-849},
url = {https://mlanthology.org/neurips/1997/vasconcelos1997neurips-multiresolution/}
}