Wavelet domain dictionary learning-based single image superresolution

Nazzal, Mahmoud; Özkaramanlı, Hüseyin

doi:10.1007/s11760-013-0602-7

Cited by 23 publications

(18 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The proposed algorithm falls into the category of wavelet domain-based SISR algorithms. Authors in [13] proposed a dictionary learning-based algorithm in the wavelet domain. The proposed algorithm learns compact dictionaries for the task of SISR.…”

Section: Related Workmentioning

confidence: 99%

“…In the wavelet-based SISR approaches [13][14][15][16], the main point to note is that they assume the LR image as the level-1 approximation image of the wavelet decomposition. Here, to recover the HR image, the task is to estimate the wavelet sub-band images representing this approximation image, and finally doing one-level inverse wavelet transform.…”

Section: Related Workmentioning

confidence: 99%

“…However, the algorithms proposed so far are unable to achieve better performance for higher scale-ups. The proposed algorithm is a wavelet domain-based algorithm inspired by the category of the SISR algorithms in the wavelet domain [13][14][15][16][17]. Most of these algorithms give state-of-the-art results.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Wavelet-Integrated Deep Networks for Single Image Super-Resolution

et al. 2019

View full text Add to dashboard Cite

We propose a scale-invariant deep neural network model based on wavelets for single image super-resolution (SISR). The wavelet approximation images and their corresponding wavelet sub-bands across all predefined scale factors are combined to form a big training data set. Then, mappings are determined between the wavelet sub-band images and their corresponding approximation images. Finally, the gradient clipping process is used to boost the training speed of the algorithm. Furthermore, stationary wavelet transform (SWT) is used instead of a discrete wavelet transform (DWT), due to its up-scaling property. In this way, we can preserve more information about the images. In the proposed model, the high-resolution image is recovered with detailed features, due to redundancy (across the scale) property of wavelets. Experimental results show that the proposed model outperforms state-of-the algorithms in terms of peak signal-to-noise ratio (PSNR) and structural similarity index measure (SSIM).

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Wavelet-Integrated Deep Networks for Single Image Super-Resolution

et al. 2019

View full text Add to dashboard Cite

show abstract

“…The other more recent and applicable approach learns a dictionary based on natural signal samples obtained from real-life example signals. In this way, a first estimate of the dictionary is established, and each sample signal helps further refine the dictionary [15]. There have been lots of motivated researchers devoted themselves to conduct research in this area.…”

Section: Introductionmentioning

confidence: 99%

A novel regularized K-SVD dictionary learning based medical image super-resolution algorithm

Yang

Zhang

Peng

et al. 2015

Multimed Tools Appl

View full text Add to dashboard Cite

Recently sparse representations over learned dictionaries have been proven to be a very successful representation method for many image processing applications. In the medical image processing community, image super-resolution has been playing a vital role to make the computer based diagnosis more efficient and accurate. Resolution enhancement through conventional interpolation methods strongly affects the precision of consequent processing steps such as segmentation and registration. In this paper, we propose a novel regularized K-SVD dictionary learning based medical image super-resolution algorithm. First, the dictionary is trained using the modified version of the K-SVD dictionary learning procedure. The sparse coding phase of the K-SVD dictionary learning scheme is then enhanced incorporating a simple and an efficient regularized version of orthogonal matching pursuit. In addition, the dictionary update stage allows for an arbitrary number of atoms at the same time and sparse coefficient vector. In the SR reconstruction procedure, ROMP is adopted to find out for the vector of sparse representation coefficients for the underlying patch. In the final part, mathematical optimization finalizes the SR work effectively. The numerical analysis and experimental simulation prove the feasibility and robustness of our proposed methodology compared with other state-of-the-art algorithms.

show abstract

“…Zeyde and Elad improved the Yang's algorithm, who used orthogonal matching pursuit (OMP) algorithm to reduce the computational complexity [3,4,5]. In 2014, Nazzal put forward an algorithm based on learned dictionaries in the wavelet domain, and further improved the reconstruction quality and speed of Zeyde's algorithm [6]. These researches mainly used common images as the processing data.…”

Section: Introductionmentioning

confidence: 99%

MODIS images super-resolution algorithm via sparse representation

Pang

Ren

et al. 2014

SPIE Proceedings

View full text Add to dashboard Cite

Based on the current mainstream algorithms, an effective super-resolution algorithm via sparse representation for MODIS remote sensing images is proposed in the paper. The basic idea behind the proposed algorithm is to obtain the redundant dictionaries deriving from high-resolution Landsat ETM+ images and low-resolution MODIS images, further give the instruction for reconstructing high-resolution MODIS images. Feature extraction is one vital part included in the procedure of dictionary training. The features are extracted from the wavelet-domain images as training samples, and then more effective dictionaries for high-resolution image reconstruction are obtained by applying the k-singular value decomposition (K-SVD) dictionary training algorithm. The experimental results demonstrate the proposed algorithm improved the reconstruction quality both visually and quantitatively. Compared with the traditional algorithm, the PSNR value approximately increases by 1.1 dB and SSIM value increases by 0.07. Moreover, both the quality and computational efficiency of the proposed algorithm can be improved given the appropriate number of atoms.

show abstract

Wavelet domain dictionary learning-based single image superresolution

Cited by 23 publications

References 22 publications

Wavelet-Integrated Deep Networks for Single Image Super-Resolution

Wavelet-Integrated Deep Networks for Single Image Super-Resolution

A novel regularized K-SVD dictionary learning based medical image super-resolution algorithm

MODIS images super-resolution algorithm via sparse representation

Contact Info

Product

Resources

About