Split Bregman algorithms for sparse / joint-sparse and low-rank signal recovery: Application in compressive hyperspectral imaging

Gogna, Anupriya; Shukla, Ankita; Agarwal, Himanish; Majumdar, Angshul

doi:10.1109/icip.2014.7025260

Cited by 36 publications

(26 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This approach had been very successful in solving multiple penalty optimization problems [10,11]. We repeat synthesis prior problem for the sake of convenience.…”

Section: Proposed Algorithmsmentioning

confidence: 99%

Mixed Gaussian and impulse denoising of hyperspectral images

Aggarwal

Majumdar

2015

2015 IEEE International Geoscience and Remote Sensing Symposium (IGARSS)

View full text Add to dashboard Cite

Hyperspectral image denoising is an important preprocessing step in the analysis of hyperspectral images in several applicaitons domains. These images often gets corrupted by various kinds of noise during acquisition process. There are several studies on reducing Gaussian noise from hyperspectral images. This work addresses the problem of reducing mixed noise from hyperspectral images; in particular a mixture of Gaussian and impulse noise has been considered. The proposed image acquisition model explicitly accounts for both Gaussian and impulse noise as additive noise. This mixed noise reduction problem has been formulated as synthesis prior optimization problem which exploits inherent spatio-spectral correlation present in hyperspectral images. Split-Bregman based approach has been utilized to solve resulting optimization problem. Experiements were conducted using both synthetic noise as well as real noisy hyperspectral images. Experimental results have been quantified using peak signal to noise ratio (PSNR) and structural similarity index (SSIM). A comparative study with an existing low-rank based image denoising approaches has also been carried out. Both quantitative and qualitative results suggest the superiority of proposed approach.

show abstract

“…This approach had been very successful in solving multiple penalty optimization problems [10,11]. We repeat synthesis prior problem for the sake of convenience.…”

Section: Proposed Algorithmsmentioning

confidence: 99%

Mixed Gaussian and impulse denoising of hyperspectral images

Aggarwal

Majumdar

2015

2015 IEEE International Geoscience and Remote Sensing Symposium (IGARSS)

View full text Add to dashboard Cite

show abstract

“…The 'Bostswana' data is acquired by the NASA EO-l satellite over the Okavango Delta, Botswana in 2001 I. We choose Indian Pine and Botswana data because they 462 are widely used in the researches of HSI compressive sensing, sparse representation and classification [l7] [18] [19] .…”

Section: A Simulation Setupmentioning

confidence: 99%

Fast holo-kronecker compressive sensing for hyperspectral image

Zhao

Wang

Shen

2015

2015 10th International Conference on Communications and Networking in China (ChinaCom)

View full text Add to dashboard Cite

Compressive sensing of hyperspectral image (HSI) faces the difficulties of complex computation and much information redundancies. In this paper, we propose a highly-efficient compressive sensing framework including sampling method and its corresponding reconstruction algorithm for HSI. Kronecker product is used to generate the sparsifying basis and measurement matrices. Both the data in spatial dimensions and spectral dimension are compressed, resulting an enhanced sampling efficiency. Very few measurements are needed for a successful reconstruction. We combine the sparsity model and low multilinear-rank model for fast and accurate reconstruction. Iterative algorithm is employed to reconstruct the data only in one dimension of HSI independently instead of all dimensions globally, which can speed up the reconstruction.

show abstract

“…But, Gaussian noise removal from such images is a well studied topic. In fact there are several studies which study the problem of compressive hyper-spectral imaging in the presence of Gaussian noise [9,10]. These studies recover the image by solving the following optimization problem:…”

Section: Literature Reviewmentioning

confidence: 99%

“…To tune the second parameter �, we fix AI to the obtained value and again use the L-curve method to determine 1.. 2. Such a technique, although sub-optimal have showed good results in practice before [10,14,15]. We did not fine tune the parameters and varied them on the log scale (100,10, 1,0.1 etc.).…”

Section: Experimental Ev Alua Nonmentioning

confidence: 99%

“…One can apply these techniques [4][5][6][7][8] on each of the spectral bands (of the hyper-sepctral datacube) separately. But such an approach may not yield the best results; since it would not account for the inter-band spectral correlation, Studies in compressive multi-spectral imaging [9][10][11] have shown that the images in different spectral bands look very similar to each other and consequently have a common sparse Uoint-sparsity) support in certain transform domains. Following these studies, we will exploit the common sparse support of the hyper-spectral images while denoising.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Hyper-spectral impulse denoising: A row-sparse Blind Compressed Sensing formulation

Majumdar

Ansari

Aggarwal

2015

2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

This paper addresses the problem of impulse denoising from hyper-spectral images. Impulse noise is sparse; removing impulse noise requires minimizing an Irnorm data fidelity term. Prior studies have exploited the intra band spatial correlation (leading to sparsity in transform domain) and inter-band spectral-correlation Uoint sparsity) of hyper-spectral images for Gaussian denoising. In this work, we propose to learn the joint-sparsity promoting dictionary adaptively from the data for impulse denoising problems. Unlike dictionary learning techniques, the sparsifying dictionary is not learnt in an oftline training phase. We follow the Blind Compressed Sensing (BCS) framework -dictionary learning and denoising proceeds simultaneously. The optimization problem that arises out of our formulation is solved using the Split Bregman approach. The proposed algorithm, when compared against prior techniques (on real hyper spectral datasets) shows more than 5dB improvement in PSNR on average.

show abstract

Split Bregman algorithms for sparse / joint-sparse and low-rank signal recovery: Application in compressive hyperspectral imaging

Cited by 36 publications

References 23 publications

Mixed Gaussian and impulse denoising of hyperspectral images

Mixed Gaussian and impulse denoising of hyperspectral images

Fast holo-kronecker compressive sensing for hyperspectral image

Hyper-spectral impulse denoising: A row-sparse Blind Compressed Sensing formulation

Contact Info

Product

Resources

About