Multilevel dictionary learning for sparse representation of images

Jayaraman J. Thiagarajan, Karthikeyan N. Ramamurthy, Andreas Spanias

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

31 Citations (Scopus)

Abstract

Adaptive data-driven dictionaries for sparse approximations provide superior performance compared to predefined dictionaries in applications involving representation and classification of data. In this paper, we propose a novel algorithm for learning global dictionaries particularly suited to the sparse representation of natural images. The proposed algorithm uses a hierarchical energy based learning approach to learn a multilevel dictionary. The atoms that contribute the most energy to the representation are learned in the first level and those that contribute lesser energies are learned in the subsequent levels. The learned multilevel dictionary is compared to a dictionary learned using the K-SVD algorithm. Reconstruction results using a small number of non-zero coefficients demonstrate the advantage of exploiting energy hierarchy using multilevel dictionaries, pointing to potential applications in low bit-rate image compression. Superior performance in compressed sensing using optimized sensing matrices with small number of measurements is also demonstrated.

Original languageEnglish (US)
Title of host publication2011 Digital Signal Processing and Signal Processing Education Meeting, DSP/SPE 2011 - Proceedings
Pages271-276
Number of pages6
DOIs
StatePublished - 2011
Event2011 Digital Signal Processing and Signal Processing Education Meeting, DSP/SPE 2011 - Sedona, AZ, United States
Duration: Jan 4 2011Jan 7 2011

Other

Other2011 Digital Signal Processing and Signal Processing Education Meeting, DSP/SPE 2011
CountryUnited States
CitySedona, AZ
Period1/4/111/7/11

Fingerprint

Glossaries
dictionary
learning
energy
Compressed sensing
Singular value decomposition
Image compression
performance
reconstruction
Atoms

Keywords

  • compressed sensing
  • dictionary learning
  • K-hyperline clustering
  • natural image statistics
  • sparse representations

ASJC Scopus subject areas

  • Computer Networks and Communications
  • Signal Processing
  • Education

Cite this

Thiagarajan, J. J., Ramamurthy, K. N., & Spanias, A. (2011). Multilevel dictionary learning for sparse representation of images. In 2011 Digital Signal Processing and Signal Processing Education Meeting, DSP/SPE 2011 - Proceedings (pp. 271-276). [5739224] https://doi.org/10.1109/DSP-SPE.2011.5739224

Multilevel dictionary learning for sparse representation of images. / Thiagarajan, Jayaraman J.; Ramamurthy, Karthikeyan N.; Spanias, Andreas.

2011 Digital Signal Processing and Signal Processing Education Meeting, DSP/SPE 2011 - Proceedings. 2011. p. 271-276 5739224.

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

Thiagarajan, JJ, Ramamurthy, KN & Spanias, A 2011, Multilevel dictionary learning for sparse representation of images. in 2011 Digital Signal Processing and Signal Processing Education Meeting, DSP/SPE 2011 - Proceedings., 5739224, pp. 271-276, 2011 Digital Signal Processing and Signal Processing Education Meeting, DSP/SPE 2011, Sedona, AZ, United States, 1/4/11. https://doi.org/10.1109/DSP-SPE.2011.5739224
Thiagarajan JJ, Ramamurthy KN, Spanias A. Multilevel dictionary learning for sparse representation of images. In 2011 Digital Signal Processing and Signal Processing Education Meeting, DSP/SPE 2011 - Proceedings. 2011. p. 271-276. 5739224 https://doi.org/10.1109/DSP-SPE.2011.5739224
Thiagarajan, Jayaraman J. ; Ramamurthy, Karthikeyan N. ; Spanias, Andreas. / Multilevel dictionary learning for sparse representation of images. 2011 Digital Signal Processing and Signal Processing Education Meeting, DSP/SPE 2011 - Proceedings. 2011. pp. 271-276
@inproceedings{3d2aa6495dda406bb27c1d6547fe7992,
title = "Multilevel dictionary learning for sparse representation of images",
abstract = "Adaptive data-driven dictionaries for sparse approximations provide superior performance compared to predefined dictionaries in applications involving representation and classification of data. In this paper, we propose a novel algorithm for learning global dictionaries particularly suited to the sparse representation of natural images. The proposed algorithm uses a hierarchical energy based learning approach to learn a multilevel dictionary. The atoms that contribute the most energy to the representation are learned in the first level and those that contribute lesser energies are learned in the subsequent levels. The learned multilevel dictionary is compared to a dictionary learned using the K-SVD algorithm. Reconstruction results using a small number of non-zero coefficients demonstrate the advantage of exploiting energy hierarchy using multilevel dictionaries, pointing to potential applications in low bit-rate image compression. Superior performance in compressed sensing using optimized sensing matrices with small number of measurements is also demonstrated.",
keywords = "compressed sensing, dictionary learning, K-hyperline clustering, natural image statistics, sparse representations",
author = "Thiagarajan, {Jayaraman J.} and Ramamurthy, {Karthikeyan N.} and Andreas Spanias",
year = "2011",
doi = "10.1109/DSP-SPE.2011.5739224",
language = "English (US)",
isbn = "9781612842271",
pages = "271--276",
booktitle = "2011 Digital Signal Processing and Signal Processing Education Meeting, DSP/SPE 2011 - Proceedings",

}

TY - GEN

T1 - Multilevel dictionary learning for sparse representation of images

AU - Thiagarajan, Jayaraman J.

AU - Ramamurthy, Karthikeyan N.

AU - Spanias, Andreas

PY - 2011

Y1 - 2011

N2 - Adaptive data-driven dictionaries for sparse approximations provide superior performance compared to predefined dictionaries in applications involving representation and classification of data. In this paper, we propose a novel algorithm for learning global dictionaries particularly suited to the sparse representation of natural images. The proposed algorithm uses a hierarchical energy based learning approach to learn a multilevel dictionary. The atoms that contribute the most energy to the representation are learned in the first level and those that contribute lesser energies are learned in the subsequent levels. The learned multilevel dictionary is compared to a dictionary learned using the K-SVD algorithm. Reconstruction results using a small number of non-zero coefficients demonstrate the advantage of exploiting energy hierarchy using multilevel dictionaries, pointing to potential applications in low bit-rate image compression. Superior performance in compressed sensing using optimized sensing matrices with small number of measurements is also demonstrated.

AB - Adaptive data-driven dictionaries for sparse approximations provide superior performance compared to predefined dictionaries in applications involving representation and classification of data. In this paper, we propose a novel algorithm for learning global dictionaries particularly suited to the sparse representation of natural images. The proposed algorithm uses a hierarchical energy based learning approach to learn a multilevel dictionary. The atoms that contribute the most energy to the representation are learned in the first level and those that contribute lesser energies are learned in the subsequent levels. The learned multilevel dictionary is compared to a dictionary learned using the K-SVD algorithm. Reconstruction results using a small number of non-zero coefficients demonstrate the advantage of exploiting energy hierarchy using multilevel dictionaries, pointing to potential applications in low bit-rate image compression. Superior performance in compressed sensing using optimized sensing matrices with small number of measurements is also demonstrated.

KW - compressed sensing

KW - dictionary learning

KW - K-hyperline clustering

KW - natural image statistics

KW - sparse representations

UR - http://www.scopus.com/inward/record.url?scp=79954563095&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79954563095&partnerID=8YFLogxK

U2 - 10.1109/DSP-SPE.2011.5739224

DO - 10.1109/DSP-SPE.2011.5739224

M3 - Conference contribution

AN - SCOPUS:79954563095

SN - 9781612842271

SP - 271

EP - 276

BT - 2011 Digital Signal Processing and Signal Processing Education Meeting, DSP/SPE 2011 - Proceedings

ER -