Multiple Model Fitting by Evolutionary Dynamics

Michael Donoser, Martin Hirzer, Dieter Schmalstieg

Research output: Chapter in Book/Report/Conference proceedingConference contributionResearchpeer-review

Abstract

We propose a novel multiple model fitting method based on outlier insensitive evolutionary dynamics, fulfilling several important requirements. Our method automatically identifies a unspecified number of models and is robust to noise and outliers in the data. Furthermore, we are able to handle overlapping models, by allowing that data points are assigned to more than one model. This is implicitly handled during model fitting and not as a post-processing step. Gross outliers are directly identified, by letting some points unassigned. We also introduce a technique, considering nearest neighbor analysis, to significantly reduce computation time, while maintaining model fitting accuracy. We show experiments on real-world and synthetic data, achieving accurate model fitting results also demonstrating an application of plane fitting on a consumer hardware providing RGB-D video streams.
Original languageEnglish
Title of host publicationProceedings of the International Conference on Pattern Recognition (ICPR)
Publisher.
Pages3816-3821
DOIs
Publication statusPublished - 2014
EventInternational Conference on Pattern Recognition - Stockholm, Sweden
Duration: 24 Aug 201428 Aug 2014

Conference

ConferenceInternational Conference on Pattern Recognition
CountrySweden
CityStockholm
Period24/08/1428/08/14

Fingerprint

Hardware
Processing
Experiments

Fields of Expertise

  • Information, Communication & Computing

Cite this

Donoser, M., Hirzer, M., & Schmalstieg, D. (2014). Multiple Model Fitting by Evolutionary Dynamics. In Proceedings of the International Conference on Pattern Recognition (ICPR) (pp. 3816-3821). .. https://doi.org/10.1109/ICPR.2014.655

Multiple Model Fitting by Evolutionary Dynamics. / Donoser, Michael; Hirzer, Martin; Schmalstieg, Dieter.

Proceedings of the International Conference on Pattern Recognition (ICPR). ., 2014. p. 3816-3821.

Research output: Chapter in Book/Report/Conference proceedingConference contributionResearchpeer-review

Donoser, M, Hirzer, M & Schmalstieg, D 2014, Multiple Model Fitting by Evolutionary Dynamics. in Proceedings of the International Conference on Pattern Recognition (ICPR). ., pp. 3816-3821, International Conference on Pattern Recognition, Stockholm, Sweden, 24/08/14. https://doi.org/10.1109/ICPR.2014.655
Donoser M, Hirzer M, Schmalstieg D. Multiple Model Fitting by Evolutionary Dynamics. In Proceedings of the International Conference on Pattern Recognition (ICPR). . 2014. p. 3816-3821 https://doi.org/10.1109/ICPR.2014.655
Donoser, Michael ; Hirzer, Martin ; Schmalstieg, Dieter. / Multiple Model Fitting by Evolutionary Dynamics. Proceedings of the International Conference on Pattern Recognition (ICPR). ., 2014. pp. 3816-3821
@inproceedings{09816e784488415393d175b751dc104d,
title = "Multiple Model Fitting by Evolutionary Dynamics",
abstract = "We propose a novel multiple model fitting method based on outlier insensitive evolutionary dynamics, fulfilling several important requirements. Our method automatically identifies a unspecified number of models and is robust to noise and outliers in the data. Furthermore, we are able to handle overlapping models, by allowing that data points are assigned to more than one model. This is implicitly handled during model fitting and not as a post-processing step. Gross outliers are directly identified, by letting some points unassigned. We also introduce a technique, considering nearest neighbor analysis, to significantly reduce computation time, while maintaining model fitting accuracy. We show experiments on real-world and synthetic data, achieving accurate model fitting results also demonstrating an application of plane fitting on a consumer hardware providing RGB-D video streams.",
author = "Michael Donoser and Martin Hirzer and Dieter Schmalstieg",
year = "2014",
doi = "10.1109/ICPR.2014.655",
language = "English",
pages = "3816--3821",
booktitle = "Proceedings of the International Conference on Pattern Recognition (ICPR)",
publisher = ".",

}

TY - GEN

T1 - Multiple Model Fitting by Evolutionary Dynamics

AU - Donoser, Michael

AU - Hirzer, Martin

AU - Schmalstieg, Dieter

PY - 2014

Y1 - 2014

N2 - We propose a novel multiple model fitting method based on outlier insensitive evolutionary dynamics, fulfilling several important requirements. Our method automatically identifies a unspecified number of models and is robust to noise and outliers in the data. Furthermore, we are able to handle overlapping models, by allowing that data points are assigned to more than one model. This is implicitly handled during model fitting and not as a post-processing step. Gross outliers are directly identified, by letting some points unassigned. We also introduce a technique, considering nearest neighbor analysis, to significantly reduce computation time, while maintaining model fitting accuracy. We show experiments on real-world and synthetic data, achieving accurate model fitting results also demonstrating an application of plane fitting on a consumer hardware providing RGB-D video streams.

AB - We propose a novel multiple model fitting method based on outlier insensitive evolutionary dynamics, fulfilling several important requirements. Our method automatically identifies a unspecified number of models and is robust to noise and outliers in the data. Furthermore, we are able to handle overlapping models, by allowing that data points are assigned to more than one model. This is implicitly handled during model fitting and not as a post-processing step. Gross outliers are directly identified, by letting some points unassigned. We also introduce a technique, considering nearest neighbor analysis, to significantly reduce computation time, while maintaining model fitting accuracy. We show experiments on real-world and synthetic data, achieving accurate model fitting results also demonstrating an application of plane fitting on a consumer hardware providing RGB-D video streams.

U2 - 10.1109/ICPR.2014.655

DO - 10.1109/ICPR.2014.655

M3 - Conference contribution

SP - 3816

EP - 3821

BT - Proceedings of the International Conference on Pattern Recognition (ICPR)

PB - .

ER -