Improved sparse coding using manifold projections

Karthikeyan Natesan Ramamurthy, Jayaraman J. Thiagarajan, Andreas Spanias

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

9 Citations (Scopus)

Abstract

Sparse representations using predefined and learned dictionaries have widespread applications in signal and image processing. Sparse approximation techniques can be used to recover data from its low dimensional corrupted observations, based on the knowledge that the data is sparsely representable using a known dictionary. In this paper, we propose a method to improve data recovery by ensuring that the data recovered using sparse approximation is close its manifold. This is achieved by performing regularization using examples from the data manifold. This technique is particularly useful when the observations are highly reduced in dimensions when compared to the data and corrupted with high noise. Using an example application of image inpainting, we demonstrate that the proposed algorithm achieves a reduction in reconstruction error in comparison to using only sparse coding with predefined and learned dictionaries, when the percentage of missing pixels is high.

Original languageEnglish (US)
Title of host publicationProceedings - International Conference on Image Processing, ICIP
Pages1237-1240
Number of pages4
DOIs
StatePublished - 2011
Event2011 18th IEEE International Conference on Image Processing, ICIP 2011 - Brussels, Belgium
Duration: Sep 11 2011Sep 14 2011

Other

Other2011 18th IEEE International Conference on Image Processing, ICIP 2011
CountryBelgium
CityBrussels
Period9/11/119/14/11

Fingerprint

Glossaries
Signal processing
Image processing
Pixels
Recovery

Keywords

  • dictionary learning
  • image inpainting
  • manifold projection
  • sparse representation

ASJC Scopus subject areas

  • Software
  • Computer Vision and Pattern Recognition
  • Signal Processing

Cite this

Ramamurthy, K. N., Thiagarajan, J. J., & Spanias, A. (2011). Improved sparse coding using manifold projections. In Proceedings - International Conference on Image Processing, ICIP (pp. 1237-1240). [6115656] https://doi.org/10.1109/ICIP.2011.6115656

Improved sparse coding using manifold projections. / Ramamurthy, Karthikeyan Natesan; Thiagarajan, Jayaraman J.; Spanias, Andreas.

Proceedings - International Conference on Image Processing, ICIP. 2011. p. 1237-1240 6115656.

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

Ramamurthy, KN, Thiagarajan, JJ & Spanias, A 2011, Improved sparse coding using manifold projections. in Proceedings - International Conference on Image Processing, ICIP., 6115656, pp. 1237-1240, 2011 18th IEEE International Conference on Image Processing, ICIP 2011, Brussels, Belgium, 9/11/11. https://doi.org/10.1109/ICIP.2011.6115656
Ramamurthy KN, Thiagarajan JJ, Spanias A. Improved sparse coding using manifold projections. In Proceedings - International Conference on Image Processing, ICIP. 2011. p. 1237-1240. 6115656 https://doi.org/10.1109/ICIP.2011.6115656
Ramamurthy, Karthikeyan Natesan ; Thiagarajan, Jayaraman J. ; Spanias, Andreas. / Improved sparse coding using manifold projections. Proceedings - International Conference on Image Processing, ICIP. 2011. pp. 1237-1240
@inproceedings{387278aaae044bed8339a8c6329dcb6d,
title = "Improved sparse coding using manifold projections",
abstract = "Sparse representations using predefined and learned dictionaries have widespread applications in signal and image processing. Sparse approximation techniques can be used to recover data from its low dimensional corrupted observations, based on the knowledge that the data is sparsely representable using a known dictionary. In this paper, we propose a method to improve data recovery by ensuring that the data recovered using sparse approximation is close its manifold. This is achieved by performing regularization using examples from the data manifold. This technique is particularly useful when the observations are highly reduced in dimensions when compared to the data and corrupted with high noise. Using an example application of image inpainting, we demonstrate that the proposed algorithm achieves a reduction in reconstruction error in comparison to using only sparse coding with predefined and learned dictionaries, when the percentage of missing pixels is high.",
keywords = "dictionary learning, image inpainting, manifold projection, sparse representation",
author = "Ramamurthy, {Karthikeyan Natesan} and Thiagarajan, {Jayaraman J.} and Andreas Spanias",
year = "2011",
doi = "10.1109/ICIP.2011.6115656",
language = "English (US)",
isbn = "9781457713033",
pages = "1237--1240",
booktitle = "Proceedings - International Conference on Image Processing, ICIP",

}

TY - GEN

T1 - Improved sparse coding using manifold projections

AU - Ramamurthy, Karthikeyan Natesan

AU - Thiagarajan, Jayaraman J.

AU - Spanias, Andreas

PY - 2011

Y1 - 2011

N2 - Sparse representations using predefined and learned dictionaries have widespread applications in signal and image processing. Sparse approximation techniques can be used to recover data from its low dimensional corrupted observations, based on the knowledge that the data is sparsely representable using a known dictionary. In this paper, we propose a method to improve data recovery by ensuring that the data recovered using sparse approximation is close its manifold. This is achieved by performing regularization using examples from the data manifold. This technique is particularly useful when the observations are highly reduced in dimensions when compared to the data and corrupted with high noise. Using an example application of image inpainting, we demonstrate that the proposed algorithm achieves a reduction in reconstruction error in comparison to using only sparse coding with predefined and learned dictionaries, when the percentage of missing pixels is high.

AB - Sparse representations using predefined and learned dictionaries have widespread applications in signal and image processing. Sparse approximation techniques can be used to recover data from its low dimensional corrupted observations, based on the knowledge that the data is sparsely representable using a known dictionary. In this paper, we propose a method to improve data recovery by ensuring that the data recovered using sparse approximation is close its manifold. This is achieved by performing regularization using examples from the data manifold. This technique is particularly useful when the observations are highly reduced in dimensions when compared to the data and corrupted with high noise. Using an example application of image inpainting, we demonstrate that the proposed algorithm achieves a reduction in reconstruction error in comparison to using only sparse coding with predefined and learned dictionaries, when the percentage of missing pixels is high.

KW - dictionary learning

KW - image inpainting

KW - manifold projection

KW - sparse representation

UR - http://www.scopus.com/inward/record.url?scp=84856283162&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84856283162&partnerID=8YFLogxK

U2 - 10.1109/ICIP.2011.6115656

DO - 10.1109/ICIP.2011.6115656

M3 - Conference contribution

AN - SCOPUS:84856283162

SN - 9781457713033

SP - 1237

EP - 1240

BT - Proceedings - International Conference on Image Processing, ICIP

ER -