Pixel-to-Pixel Matching for Image Recognition Using Hungarian Graph Matching

DMCA

Pixel-to-Pixel Matching for Image Recognition Using Hungarian Graph Matching (2004)

Cached

Download Links

Other Repositories/Bibliography

DBLP

by Daniel Keysers , Thomas Deselaers , Hermann Ney

Venue:	In DAGM 2004, Pattern Recognition, 26th DAGM Symposium
Citations:	10 - 6 self

BibTeX

@INPROCEEDINGS{Keysers04pixel-to-pixelmatching,
    author = {Daniel Keysers and Thomas Deselaers and Hermann Ney},
    title = {Pixel-to-Pixel Matching for Image Recognition Using Hungarian Graph Matching},
    booktitle = {In DAGM 2004, Pattern Recognition, 26th DAGM Symposium},
    year = {2004},
    pages = {154--162}
}

OpenURL

Abstract

A fundamental problem in image recognition is to evaluate the similarity of two images. This can be done by searching for the best pixel-to-pixel matching taking into account suitable constraints. In this paper, we present an extension of a zero-order matching model called the image distortion model that yields state-of-the-art classification results for di#erent tasks. We include the constraint that in the matching process each pixel of both compared images must be matched at least once. The optimal matching under this constraint can be determined using the Hungarian algorithm. The additional constraint leads to more homogeneous displacement fields in the matching. The method reduces the error rate of a nearest neighbor classifier on the well known USPS handwritten digit recognition task from 2.4% to 2.2%.

Keyphrases

pixel-to-pixel matching account suitable constraint optimal matching di erent task nearest neighbor classifier additional constraint usps handwritten digit recognition task matching process fundamental problem image distortion model homogeneous displacement field hungarian algorithm state-of-the-art classification result error rate image recognition

DMCA