Abstract

Image denoising is an important technique of practical significance which serves as a preliminary step for other analyses like image feature extraction, image classification etc. Two novel methods for denoising images are proposed which deal with the case when there is no noise-free training data. The basic method consists of several phases: the first phase involves preprocessing the given noisy data matrix to obtain a good approximation matrix; the second phase involves implementing kernel principal component analysis (KPCA) on the approximation matrix obtained from the first phase. KPCA is one of the useful non-linear techniques applied to image denoising. However, an important problem faced in KPCA is estimating the denoised pre-image. Consequently, we generate a pre-image by solving a regularized regression problem. The second method is an ensemble version of the basic method that provides robustness to noisy instances. Some of the attractive properties of the proposed methods include numerical stability and ease of implementation. Also our methods are based on linear algebra and avoid any nonlinear optimization. Our methods are demonstrated on high-noise cases (for both Gaussian noise and "salt and pepper" noise) for the USPS digits dataset, and they perform better than existing alternatives both in terms of low mean square error and better visual quality of the reconstructed pre-images.

Original languageEnglish (US)
Title of host publicationProceedings - Applied Imagery Pattern Recognition Workshop
DOIs
StatePublished - 2011
Event2011 IEEE Applied Imagery Pattern Recognition Workshop: Imaging for Decision Making, AIPR 2011 - Washington, DC, United States
Duration: Oct 11 2011Oct 13 2011

Other

Other2011 IEEE Applied Imagery Pattern Recognition Workshop: Imaging for Decision Making, AIPR 2011
CountryUnited States
CityWashington, DC
Period10/11/1110/13/11

Fingerprint

Image denoising
Principal component analysis
Linear algebra
Image classification
Convergence of numerical methods
Mean square error
Feature extraction
Salts

ASJC Scopus subject areas

  • Engineering(all)

Cite this

Sahu, A., Runger, G., & Apley, D. (2011). Image denoising with a multi-phase kernel principal component approach and an ensemble version. In Proceedings - Applied Imagery Pattern Recognition Workshop [6176339] https://doi.org/10.1109/AIPR.2011.6176339

Image denoising with a multi-phase kernel principal component approach and an ensemble version. / Sahu, Anshuman; Runger, George; Apley, Daniel.

Proceedings - Applied Imagery Pattern Recognition Workshop. 2011. 6176339.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Sahu, A, Runger, G & Apley, D 2011, Image denoising with a multi-phase kernel principal component approach and an ensemble version. in Proceedings - Applied Imagery Pattern Recognition Workshop., 6176339, 2011 IEEE Applied Imagery Pattern Recognition Workshop: Imaging for Decision Making, AIPR 2011, Washington, DC, United States, 10/11/11. https://doi.org/10.1109/AIPR.2011.6176339
Sahu, Anshuman ; Runger, George ; Apley, Daniel. / Image denoising with a multi-phase kernel principal component approach and an ensemble version. Proceedings - Applied Imagery Pattern Recognition Workshop. 2011.
@inproceedings{c2a72eeb87e54d299815ef961c583c42,
title = "Image denoising with a multi-phase kernel principal component approach and an ensemble version",
abstract = "Image denoising is an important technique of practical significance which serves as a preliminary step for other analyses like image feature extraction, image classification etc. Two novel methods for denoising images are proposed which deal with the case when there is no noise-free training data. The basic method consists of several phases: the first phase involves preprocessing the given noisy data matrix to obtain a good approximation matrix; the second phase involves implementing kernel principal component analysis (KPCA) on the approximation matrix obtained from the first phase. KPCA is one of the useful non-linear techniques applied to image denoising. However, an important problem faced in KPCA is estimating the denoised pre-image. Consequently, we generate a pre-image by solving a regularized regression problem. The second method is an ensemble version of the basic method that provides robustness to noisy instances. Some of the attractive properties of the proposed methods include numerical stability and ease of implementation. Also our methods are based on linear algebra and avoid any nonlinear optimization. Our methods are demonstrated on high-noise cases (for both Gaussian noise and {"}salt and pepper{"} noise) for the USPS digits dataset, and they perform better than existing alternatives both in terms of low mean square error and better visual quality of the reconstructed pre-images.",
author = "Anshuman Sahu and George Runger and Daniel Apley",
year = "2011",
doi = "10.1109/AIPR.2011.6176339",
language = "English (US)",
isbn = "9781467302159",
booktitle = "Proceedings - Applied Imagery Pattern Recognition Workshop",

}

TY - GEN

T1 - Image denoising with a multi-phase kernel principal component approach and an ensemble version

AU - Sahu, Anshuman

AU - Runger, George

AU - Apley, Daniel

PY - 2011

Y1 - 2011

N2 - Image denoising is an important technique of practical significance which serves as a preliminary step for other analyses like image feature extraction, image classification etc. Two novel methods for denoising images are proposed which deal with the case when there is no noise-free training data. The basic method consists of several phases: the first phase involves preprocessing the given noisy data matrix to obtain a good approximation matrix; the second phase involves implementing kernel principal component analysis (KPCA) on the approximation matrix obtained from the first phase. KPCA is one of the useful non-linear techniques applied to image denoising. However, an important problem faced in KPCA is estimating the denoised pre-image. Consequently, we generate a pre-image by solving a regularized regression problem. The second method is an ensemble version of the basic method that provides robustness to noisy instances. Some of the attractive properties of the proposed methods include numerical stability and ease of implementation. Also our methods are based on linear algebra and avoid any nonlinear optimization. Our methods are demonstrated on high-noise cases (for both Gaussian noise and "salt and pepper" noise) for the USPS digits dataset, and they perform better than existing alternatives both in terms of low mean square error and better visual quality of the reconstructed pre-images.

AB - Image denoising is an important technique of practical significance which serves as a preliminary step for other analyses like image feature extraction, image classification etc. Two novel methods for denoising images are proposed which deal with the case when there is no noise-free training data. The basic method consists of several phases: the first phase involves preprocessing the given noisy data matrix to obtain a good approximation matrix; the second phase involves implementing kernel principal component analysis (KPCA) on the approximation matrix obtained from the first phase. KPCA is one of the useful non-linear techniques applied to image denoising. However, an important problem faced in KPCA is estimating the denoised pre-image. Consequently, we generate a pre-image by solving a regularized regression problem. The second method is an ensemble version of the basic method that provides robustness to noisy instances. Some of the attractive properties of the proposed methods include numerical stability and ease of implementation. Also our methods are based on linear algebra and avoid any nonlinear optimization. Our methods are demonstrated on high-noise cases (for both Gaussian noise and "salt and pepper" noise) for the USPS digits dataset, and they perform better than existing alternatives both in terms of low mean square error and better visual quality of the reconstructed pre-images.

UR - http://www.scopus.com/inward/record.url?scp=84860701746&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84860701746&partnerID=8YFLogxK

U2 - 10.1109/AIPR.2011.6176339

DO - 10.1109/AIPR.2011.6176339

M3 - Conference contribution

SN - 9781467302159

BT - Proceedings - Applied Imagery Pattern Recognition Workshop

ER -