Single Image Super-Resolution via Sparse Representation over Directionally Structured Dictionaries Based on the Patch Gradient Phase Angle

Size: px
Start display at page:

Download "Single Image Super-Resolution via Sparse Representation over Directionally Structured Dictionaries Based on the Patch Gradient Phase Angle"

Transcription

1 2014 UKSim-AMSS 8th European Modelling Symposium Single Image Super-Resolution via Sparse Representation over Directionally Structured Dictionaries Based on the Patch Gradient Phase Angle Mahmoud Nazzal, Faezeh Yeganli and Huseyin Ozkaramanli Electrical and Electronic Engineering Department Eastern Mediterranean University Gazimagusa, via Mersin 10, Turkey Abstract We propose a single-image super-resolution algorithm based on sparse representation over a set of cluster dictionary pairs. For each cluster, a directionally structured dictionary pair is designed. The dominant angle in the patch gradient phase matrix is employed as an approximately scaleinvariant measure. This measure serves for patch clustering and sparse model selection. The dominant phase angle of each low resolution patch is found and used to identify its corresponding cluster. Then, the sparse coding coefficients of this patch with respect to the low resolution cluster dictionary are calculated. These coefficients are imposed on the high resolution dictionary of the same cluster to obtain a high resolution patch estimate. In experiments conducted on several images, the proposed algorithm is shown to outperform the algorithm that uses a single universal dictionary pair, and to be competitive to the state-of-the art algorithm. This is validated in terms of PSNR, SSIM and visual comparison. Keywords single-image super-resolution, sparse representation, cluster dictionary pairs, gradient operator. I. INTRODUCTION Singe-image super-resolution (SR) is the process of reconstructing a high resolution (HR) image (I H ) from a low resolution (LR) image (I L ) corresponding to the same scene. The relationship between I L and I H is typically characterized by a blurring and downsampling operation, as described in (1). I L ΨI H, (1) where Ψ is the blurring and downsampling operator. It is well-known that this process is an ill-posed inverse problem. Approaches to SR can be classified into three main categories. First, interpolation methods which estimate the unknown pixel values based on the known ones. Several advanced interpolation techniques have been proposed to over-come the low-pass nature of the interpolation process. Second is the reconstruction-based methods, where constrains are applied and exploited in reconstructing the HR image. The third category is the learning-based approaches where a training process is carried out and then used during a testing process. The fundamental idea in these approaches is to utilize the correspondence between the training and testing sets. One of the most successful learning-based approaches is the sparse representationbased approach. In this framework, dictionaries are trained over example signals and then used in the testing process. In the context of sparse representation-based SR, sparsity of the representation of LR and HR image patches is used as a natural image prior to regularize the SR problem. This usage allows for assuming that the sparse coding coefficients of an image patch are scale-invariant as proposed by Yang et al. [1], [2]. Another prior that has been successfully employed in regard is the gradient operator. This operator has been employed as a natural image prior to solve ill-posed image processing problems such as SR [3], [4] and denoising [5]. The early usage of the gradient as a prior was proposed by Rudin et al. [6] in the total variation denoising framework. Another line of research considered approximating the gradient distribution such as [7] and [8]. More recent works have considered the invariance or preservation of the gradient histogram and applied it as a prior. An example is the work conducted by Zuo et al. [5] in image denoising, where they enforced the histogram of the denoised image to be similar to that of the noisy one. Another example is the work by Sun et al. [3], [4] which defines a gradient profile prior that is invariant to scale, and uses it as a generic image prior for SR. In view of the success of sparse representation in various signal and image processing applications, it is well-known that its major added-benefit is the signalfitting capability due to the use of learned dictionaries. Compared to fixed basis functions, such dictionaries are proven to better fit natural signals as they are trained over examples signals [9]. Since the representation is intended to be sparse, learned dictionaries have to be overcomplete (redundant) to allow for a good representation quality. However, high redundancy has its own drawbacks. It has been shown that having high dictionary redundancy tends to cause instability of the representation and therefore to degrade the representation quality [10], [11]. In other words, a good dictionary needs to possess a good capability in representing signals at a reasonable /14 $ IEEE DOI /EMS

2 degree of redundancy. Motivated by the above findings, several attempts have recently been made towards designing compact dictionaries without sacrificing the representation power. Intuitively, the variability of signals within a class is less than the general signal variability. A recent research trend considered learning class-dependent dictionaries. Examples of this trend include clustering the training and testing data as proposed by Dong et al. [12], where they applied K-means clustering for separating signals into several clusters, and learned cluster sub-dictionaries. The same clustering criterion is used to classify a signal into a cluster and use its sub-dictionary for the purpose of sparse coding. Along this line, researchers aimed at designing directionally structured dictionaries that correspond to directionally sensitive signal classes. In [13], Yang et al. employed multiple geometric cluster dictionaries. Each cluster is concerned with a certain directional structure. Sparse coding of a signal is carried out over the cluster dictionary that best fits this signal based on its structure. In [14], Yu et al. designed a structural dictionary composed of several orthogonal bases that correspond to different structures. Again, sparse coding of a signal is done by first selecting the best fitting basis (model) according to the signal s structure, and then calculating the sparse coding coefficient with respect to this basis. In this paper, we classify the training and testing signals into a set of directionally structured clusters. We employ the dominant phase angle (DPA) defined via the gradient operator as a classification criterion. We empirically show an acceptable degree of scale-invariance of the DPA as a natural image quantity. For each cluster, a directionally structured coupled dictionary pair is learned and then used for the purpose of sparse coding. We show that the designed dictionaries inherit the intended directional structure of their respective clusters. The proposed SR algorithm is shown to be superior to that algorithm of Yang et al. [2] that uses a universal dictionary pair. Besides, it is shown to be competitive to the state-of-the-art SR algorithms. This result is validated qualitatively in terms of the peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM) measures, and qualitatively in terms of visual comparisons. II. SINGLE IMAGE SUPER-RESOLUTION VIA SPARSE REPRESENTATION The sparse-representation SR algorithm of Yang et al. [2] is based on using patches of a LR image to reconstruct the underlying patches in the unknown HR image. Sparsity of the representation is applied as a generic natural image property that is invariant with respect to scale. Their work is based on applying two constraints on the reconstructed HR image. First is a reconstruction constraint enforcing the blurred and downsampled version of the HR image estimate to be consistent with the given LR image. Second, is a sparsity constraint which assumes that the sparse coding coefficients of a LR patch are similar to those of the underlying HR one, or scale-invariant in other words. In [1], the authors used a set of sampled patch pairs as a dictionary for sparse coding. However, this paradigm is very slow in practice. Therefore, in [2], Yang et al. employed a pair of coupled dictionaries learned from such patch pairs. They first calculated the sparse coding coefficients of a LR patch with respect to the LR dictionary. Then, they imposed these coefficients on the HR dictionary to find a HR patch estimate. For the sake of local consistency of reconstructed HR patches, it is advantageous to divide the LR image into overlapping patches [1], [2]. Correspondingly, the reconstructed HR patches are also overlapped. They are then merged at the overlap locations to generate a HR image estimate. The LR and HR dictionaries are simultaneously learned in a coupled manner. Patches of HR training images are extracted and column-stacked to form an array of HR training patches. The mean value of each patch is subtracted to allow for a better training. At the same time, LR images are obtained by applying a blurring and downsampling operation of the HR ones. Then, LR images are upsampled to the so called middle resolution (MR) to allow for better feature extraction.. Afterwards, first and second order gradient filters are operated over the MR images to extract the features. The next step is to combine the extracted features of each MR patch into a single column, and then to combine feature vectors column-wise to obtain the corresponding LR training patch array. Eventually, LR and HR training patches are used to train for a pair of coupled LR and HR dictionaries, respectively. This coupling is vital for the validity of the sparse coding coefficient invariance assumption. Given a HR vector patch x H, one may find the sparse coding coefficient vector α H of this patch over a dictionary in the same resolution level D H. Vector selection techniques such the least absolute shrinkage and selection operator (LASSO) [15] can be applied for this purpose. A sparse approximation of x H can be written as x H D H α H, (2) Analogously, one may obtain a sparse approximation for the corresponding LR patch x L. This requires a sparse coding coefficient vector α L over a LR dictionary D L. This approximation can be written as x L D L α L. (3) The same blurring and downsampling operator Ψ shown in (1) can be used to relate x L and x H. With the assumption that Ψ relates also the atoms of D L and D H, one 210

3 may write x L Ψx H ΨD H α H D L α H, (4) Based on the above analysis, the following result in concluded α H α L. Consequently, a reconstruction for x H can be obtained using D H and α L, as follows x H D H α L. (5) III. THE PROPOSED SUPER-RESOLUTION ALGORITHM In this section, we introduce the DPA as an approximately scale-invariant patch measure and empirically study its behavior with respect to scale. Then, we detail the training and testing stages of the proposed algorithm. Intuitively, phase values can characterize the directional structure of an image patch. One may define a dominant phase angle (DPA) in the phase matrix that best describes the directionality. This requires quantizing the the phase matrix and establishing a histogram for the quantized angles. It is empirically found that if a certain angle value is repeated more frequently than half of the number of elements in the phase matrix, then the patch can be considered a directional patch. The directionality of this patch is characterized by that dominant angle. Otherwise, it is considered as a non-directional patch. In this work, we opt to quantize angles into values of 0,45,90 and 135. TABLE I THE NUMBER OF HR PATCHES CLASSIFIED BASED ON DPA IN EACH CLUSTER (TOP) AND DPA SCALE-INVARIANCE RATIO IN THE SAME CLUSTER (BOTTOM). THE LISTINGS OF THE CLUSTER CONTAINING THE LARGEST NUMBER OF IMAGE PATCHES ARE IN BOLD FACE. Image C 1 C 2 C 3 C 4 C 5 Barbara BSDS Buildposter Butterfly Fence Flowers Lena ppt Text Image Average A. Approximate scale-invariance of the dominant gradient phase angle measure The image gradient operator is believed to be crucial to the perception and analysis of natural images [5]. In [3], [4], Sun et al. defined the gradient profile prior as a 1-D profile of gradient magnitudes perpendicular to image structures, and studied its behavior with respect to scale. They reported that this profile follows a certain distribution which is independent of image resolution. Then they applied this prior to the problem of single-image SR. It is well-acknowledged that the phase is generally more informative than the magnitude. It seems thus advantageous to think about exploiting information in the phase of the gradient operator. One can define the phase matrix of the gradient operator based on the horizontal (G h ) and vertical (G v ) gradients as follows Φ = arctan( Gv ). (6) Gh Figure 1. Test images from left to right and top to bottom: Barbara, BSDS , Buildposter, Butterfly, Fence, Flowers, Lena, ppt3 and Text Image 1. To investigate the impact of scale on DPA, the following experiment is conducted on the images shown in Figure 1. Each image is divided into non-overlapping 8 8 patches. A LR version of each image is obtained by filtering it with a bicubic filter and downsampling it by 2 in both dimensions. Clusters C 1 through C 4 are defined corresponding to DPA values of 0,45,90 and 135 respectively, while C 5 corresponds to patches that do not have a specific directional nature. Then the DPA value of each HR and LR patch is calculated as explained earlier. The DPA value of each patch is used to cluster it into one of the clusters. The DPA invariance is defined as the ratio between the number of LR patches correctly classified into a certain cluster and the total number of HR patches in that cluster. DPA invariance ratios for the images shown in Figure 1 are listed in Table I. One can see in Table I that the DPA invariance ratios are generally greater than 50 %. DPA invariance values are high for C 5, they are still high in C 1 and C 3 as compared to the cases of C 2 211

4 C 1 C 2 C 3 C 4 C 5 Figure 2. Reshaped example atoms of HR dictionaries in C 1 through C 5. and C 4. B. Clustering and sparse model selection with the dominant gradient phase angle measure In this work, we classify the training image patch pairs into the aforementioned five clusters. Then, we use the dictionary learning method of Yang et al. [2] to train for coupled LR and HR cluster dictionary pairs. Features extracted from the LR image patches at the MR level are used to train for the LR dictionary. The DPA of each LR patch at the MR level is used to classify it into a certain cluster. Then, the corresponding HR patch and the MR extracted features are inserted to the HR and LR training sets of that cluster, respectively. Algorithm 1 outlines the main steps of the training stage. In this setting, the purpose is to design cluster dictionaries that possess the intended directional structures of their respective clusters. Algorithm 1 The Proposed Dictionary Learning Stage. 1: INPUT: HR Training Image Set. 2: OUTPUT: A Set of Directionally Structured Cluster Dictionary Pairs. 3: Obtain a LR image for each HR one by blurring and downsampling. 4: Upsample each LR image to obtain a MR image. 5: Divide the HR and MR images into patches. 6: Extract features from the MR image by filtering. 7: Divide feature images into column patches. 8: Combine HR patches to form a HR training set and MR features to form a LR training set. 9: for Each patch in the LR training set, do 10: Calculate the DPA of the corresponding patch in the MR image, and identify the cluster number. 11: Set the MR features and the HR patch to the LR and HR training sets of this cluster. 12: end for 13: Learn a pair of coupled dictionaries for each cluster. In the reconstruction stage, a LR image is first upsampled with bicubic interpolation to the MR level. The MR image is then divided into overlapping patches. Next, the DPA of each MR patch is calculated and used to identify the cluster this patch belongs to. The next step is to calculate the sparse coding coefficients of the features extracted from the MR patch over the cluster LR dictionary. The corresponding HR patch is then reconstructed by imposing these coefficients on the HR dictionary of the same cluster. Algorithm 2 The Proposed Reconstruction Stage. 1: INPUT: A LR Image, Cluster Dictionary Pairs. 2: OUTPUT: A HR Image Estimate 3: Upsample the LR image to the required resolution level (MR). 4: Extract feature images from the MR image 5: Extract patches from the feature images and group them columnwise 6: Divide the MR image into overlapping patches. 7: for Each MR patch do 8: Calculate DPA of the MR patch. 9: Determine the cluster this patch belongs to. 10: Calculate the sparse coding coefficients of the corresponding feature vector over the cluster LR dictionary. 11: Reconstruct a HR patch as the product of the cluster HR dictionary and the calculated coefficients. 12: end for 13: Obtain a HR image estimate by merging overlapping HR patches. The final step is to reshape the overplaying HR patches and merge them to form a HR image estimate. A summary of the proposed reconstruction algorithm is outlined in Algorithm 2. Since the proposed algorithm relies on designing cluster dictionaries, they need not to be highly redundant. In his work, we choose to design 600-atom dictionaries as a good compromise between computational complexity and representation quality. This value is empirically determined to meet these requirements. The sparse coding stage is the most computationally expensive stage in the sparse reconstruction framework as it relies on vector selection [6]. Therefore, employing five compact dictionaries is expected to substantially reduce the sparse coding computational complexity and thus reduce the overall SR computational complexity, as compared to the case of employing a single highly redundant dictionary. As an example, Yang et al. [2] use a 1000-atom dictionary pair. However, the proposed algorithm requires calculating the DPA value of each MR patch. In view of these observations, it can be seen that the proposed algorithm s computation complexity is comparable to that of the algorithm of Yang et al. [2]. IV. EXPERIMENTAL RESULTS The performance of the proposed algorithm is compared to bicubic interpolation, the algorithm of Yang et al. [2] which uses a single dictionary pair and the algorithms 212

5 (a) (b) (c) (d) (e) (f) (g) (h) (i) Figure 3. Super-resolution comparison of a scene in the Butterfly image. (a) Original scene, (b) Bicubic interpolation result, (c) Peleg et al. s [16] result, (d) Yang et al. s [2] result, (e) He et al. s [17] result and (d) The proposed algorithm s result. The last row shows the difference between the original scene and (g) Yang et al. s result, (h) He et al. s result and (i) The proposed algorithm s result. of Peleg et al. [16] and He et al. [17] as state-of-theart algorithms. Comparisons are conducted in terms of the PSNR and SSIM [18] measures. In accordance with the common practice, we transform a color image into the luminance and chrominance color space and apply the SR algorithm only to the luminance component. The two chrominance components are reconstructed by bicubic interpolation. Accordingly, PSNR is calculated with the luminance components, where SSIM is calculated as the average of its values between the three color components. Results are obtained by running source codes obtained from the authors of [2], [16] and [17]. It is noted that we have paid attention to make equality of the simulation parameters of the four algorithms. Besides, each algorithm is set to work with the optimal parameters suggested by its authors. In this experiment, the proposed algorithm uses a 6 6 patch size, and five cluster dictionary pairs. We use the images in the Flickr image dataset [19] as a training set. Image patches are classified based on DPA into the five clusters. Then, 40,000 patch pairs are randomly selected and used to train for a cluster dictionary pair. The algorithm of Yang et al. [2] is set to the same parameters with the same training data set to train for a 1000-atom dictionary pair. The algorithms of Peleg et al. [16] and He et al. [17] are set to work according to the default values suggested by the authors. Figure 2 shows example reshaped atoms in the HR cluster dictionaries of the proposed algorithm. The atoms of these dictionaries clearly possess the directional structure of their respective clusters. From left to right, atoms of the first four dictionaries are generally perpendicular to the directions of 0,45,90 and 135 respectively. However, atoms in the fifth dictionary have a chaotic directional in accordance with the non-directional nature of C 5. Table II lists PSNR and SSIM values of the test images shown in Figure 1 with the aforementioned approaches. The performance of the proposed algorithm is shown with dictionaries of 600, 800 and 1000 atoms, as denoted by P 600, P 800 and, P 1000 respectively. With 600-atom dictio- 213

6 TABLE II QUANTITATIVE ASSESSMENT OF BICUBIC INTERPOLATION, THE ALGORITHMS OF PELEG ET AL. [16], YANGETAL. [2], HE ETAL. [17] AND THE PROPOSED ALGORITHM IN TERMS OF PSNR (TOP) AND SSIM (BOTTOM). Image Bicubic Yang [2] Peleg [16] He [17] P 1000 P 800 P 600 Barbara BSDS Buildposter Butterfly Fence Flowers Lena ppt Text Image Average naries, the proposed algorithm outperforms the algorithm of Yang et al. [2] and Peleg et al. [16] with average PSNR improvements of 0.19 db and 0.28 db, respectively. However, it is competitive to the algorithm of He et al. [17]. The same result is concluded in terms of SSIM. It is conclusive that using more redundancy in the dictionaries of the proposed algorithm will not significantly improve the performance. Besides, the proposed algorithm is particularly superior in handling images rich of edges in different orientations. In Figure 3 we compare our approach to bicubic interpolation and the algorithms of Peleg et al. [16], Yang et al. [2] and He et al. [17]. Figure 3 (g), (h) and (i) show the differences between the ground-truth scene and its reconstructions from Yang et al., He et al. and the proposed algorithm, respectively. Bicubic interpolation result is over-smooth. The reconstructions of Peleg et al. and Yang et al. have sharper edges. The algorithm of He et al. has further sharper edges. However, the proposed algorithm s reconstruction is the best to approximate the ground-truth scene with less artifacts. This is particularly observable in the edges and details of the butterfly s wing. This is assured by observing that the difference image of the proposed algorithm is the darkest one. V. CONCLUSION A new single-image super-resolution algorithm is proposed based on sparse coding over directionally-structured cluster dictionary pairs. Clustering is done based on the dominant angle in the gradient operator phase matrix. This angle is shown as an approximately scale-invariant patch measure. The same measure is used as a sparse model selection criterion in the reconstruction phase. With compact cluster dictionaries, the proposed algorithm is shown to be superior to the algorithm of Yang et al. [2] with an average 0.19 db PSNR improvement that uses a single dictionary pair. It is also shown to be competitive with the state-of-the-art algorithms. This result is validated in terms of PSNR and SSIM as quantities metrics and visual comparison. REFERENCES [1] J. Yang, J. Wright, Y. Ma and T. Huang, Image super-resolution as sparserepresentation of raw image patches, Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2008, pp [2] J. Yang, J. Wright, T.S. Huang and Y. Ma, Image super-resolution via sparse representation, IEEE Trans. Image Process., vol. 19, no. 11, 2010, pp [3] J. Sun, Z. Xu and H.Y. Shum, Imagesuper-resolution using gradient profile prior, Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2008, pp [4] J. Sun, Z. Xu and H.Y. Shum, Gradient Profile Prior and Its Applications in Image Super-Resolution and Enhancement, IEEE Trans. Image Process., vol. 20, no. 6, 2011, pp [5] W. Zuo, L. Zhang, C. Song, D. Zhang and H. Gao, Gradient Histogram Estimation and Preservation for Texture Enhanced Image Denoising, IEEE Trans. on Image Process., vol. 23, no. 6, 2014, pp [6] L.I. Rudin, S. Osher, and E. Fatemi, Nonlinear total variation based noise removal algorithms, Physica D, vol. 60, 1992, pp [7] R. Fergus, B. Singh, A. Hertzmann, S.T. Roweis and W.T. Freeman, Removing camera shake from a single photograph, In ACM Transactions on Graphics (TOG). vol. 25, no. 3, 2006, pp [8] A. Levin, Y. Weiss, F. Durand and W. T. Freeman, Efficient marginal likelihood optimization in blind deconvolution, Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2011, pp [9] J. Mairal, F. Bach, J. Ponce and G. Sapiro, Online learning for matrix factorization and sparse coding, J. Mach. Learn. Res., vol. 11, 2010, pp [10] D. Thanou, D. I Shuman and P. Frossard, Parametric dictionary learning for graph signals, IEEE Glob. Conf. Signal Inf. Process. (GlobalSIP), 2013, pp [11] S. Mallat and Y. Guoshen, Super-resolution with sparse mixing estimators, IEEE Trans. Image Process., vol. 19, no. 11, 2010, pp [12] W. Dong, L. Zhang, G. Shi and X. Wu, Image deblurring and super-resolution by adaptive sparsedomain selection and adaptive regularization, IEEE Trans. Image Process., vol. 20, no. 7, 2011, pp [13] S. Yang, M. Wang, Y. Chen and Y. Sun, Single-image superresolution reconstruction via learned geometric dictionaries and clustered sparse coding, IEEE Trans. Image Process., vol. 21, no. 9, 2012, pp [14] G. Yu, G. Sapiro and S. Mallat, Image modeling and enhancement via structured sparse model selection, Proc. IEEE Int. Conf. Image Process. (ICIP), 2010, pp [15] R. Tibshirani, Regression shrinkage and selection via the lasso, J. Royal. Statist. Soc. B, vol. 58, 1996, pp [16] T. Peleg and M. Elad, A statistical prediction model based on sparse representations for single image super-resolution, IEEE Trans. Image Process., vol. 23, no. 6, 2014, pp [17] H. Li, H. Qi and R. Zaretzki, Beta process joint dictionary learning for coupled feature spaces with application to single image super-resolution, Proc. IEEE Conf. Comput. Vis. Pattern Recognit. (CVPR), 2013, pp [18] Z. Wang, A.C. Bovik, H.R. Sheikh and E.P. Simoncelli, Image quality assessment: From error visibility to structural similarity, IEEE Trans. Image Process., vol. 13, no. 4, 2004, pp [19] oads.htm 214

IMAGE DENOISING TO ESTIMATE THE GRADIENT HISTOGRAM PRESERVATION USING VARIOUS ALGORITHMS

IMAGE DENOISING TO ESTIMATE THE GRADIENT HISTOGRAM PRESERVATION USING VARIOUS ALGORITHMS IMAGE DENOISING TO ESTIMATE THE GRADIENT HISTOGRAM PRESERVATION USING VARIOUS ALGORITHMS P.Mahalakshmi 1, J.Muthulakshmi 2, S.Kannadhasan 3 1,2 U.G Student, 3 Assistant Professor, Department of Electronics

More information

Learning based face hallucination techniques: A survey

Learning based face hallucination techniques: A survey Vol. 3 (2014-15) pp. 37-45. : A survey Premitha Premnath K Department of Computer Science & Engineering Vidya Academy of Science & Technology Thrissur - 680501, Kerala, India (email: premithakpnath@gmail.com)

More information

Image Interpolation using Collaborative Filtering

Image Interpolation using Collaborative Filtering Image Interpolation using Collaborative Filtering 1,2 Qiang Guo, 1,2,3 Caiming Zhang *1 School of Computer Science and Technology, Shandong Economic University, Jinan, 250014, China, qguo2010@gmail.com

More information

IMAGE DENOISING USING NL-MEANS VIA SMOOTH PATCH ORDERING

IMAGE DENOISING USING NL-MEANS VIA SMOOTH PATCH ORDERING IMAGE DENOISING USING NL-MEANS VIA SMOOTH PATCH ORDERING Idan Ram, Michael Elad and Israel Cohen Department of Electrical Engineering Department of Computer Science Technion - Israel Institute of Technology

More information

Single-Image Super-Resolution Using Multihypothesis Prediction

Single-Image Super-Resolution Using Multihypothesis Prediction Single-Image Super-Resolution Using Multihypothesis Prediction Chen Chen and James E. Fowler Department of Electrical and Computer Engineering, Geosystems Research Institute (GRI) Mississippi State University,

More information

Image Deblurring Using Adaptive Sparse Domain Selection and Adaptive Regularization

Image Deblurring Using Adaptive Sparse Domain Selection and Adaptive Regularization Volume 3, No. 3, May-June 2012 International Journal of Advanced Research in Computer Science RESEARCH PAPER Available Online at www.ijarcs.info ISSN No. 0976-5697 Image Deblurring Using Adaptive Sparse

More information

Single Image Interpolation via Adaptive Non-Local Sparsity-Based Modeling

Single Image Interpolation via Adaptive Non-Local Sparsity-Based Modeling Single Image Interpolation via Adaptive Non-Local Sparsity-Based Modeling Yaniv Romano The Electrical Engineering Department Matan Protter The Computer Science Department Michael Elad The Computer Science

More information

Augmented Coupled Dictionary Learning for Image Super-Resolution

Augmented Coupled Dictionary Learning for Image Super-Resolution Augmented Coupled Dictionary Learning for Image Super-Resolution Muhammad Rushdi and Jeffrey Ho Computer and Information Science and Engineering University of Florida Gainesville, Florida, U.S.A. Email:

More information

Single image super-resolution by directionally structured coupled dictionary learning

Single image super-resolution by directionally structured coupled dictionary learning Ahmed and Shah EURASIP Journal on Image and Video Processing (2016) 2016:36 DOI 10.1186/s13640-016-0141-6 EURASIP Journal on Image and Video Processing RESEARCH Open Access Single image super-resolution

More information

Patch-Based Color Image Denoising using efficient Pixel-Wise Weighting Techniques

Patch-Based Color Image Denoising using efficient Pixel-Wise Weighting Techniques Patch-Based Color Image Denoising using efficient Pixel-Wise Weighting Techniques Syed Gilani Pasha Assistant Professor, Dept. of ECE, School of Engineering, Central University of Karnataka, Gulbarga,

More information

Image Restoration and Background Separation Using Sparse Representation Framework

Image Restoration and Background Separation Using Sparse Representation Framework Image Restoration and Background Separation Using Sparse Representation Framework Liu, Shikun Abstract In this paper, we introduce patch-based PCA denoising and k-svd dictionary learning method for the

More information

IMAGE RESTORATION VIA EFFICIENT GAUSSIAN MIXTURE MODEL LEARNING

IMAGE RESTORATION VIA EFFICIENT GAUSSIAN MIXTURE MODEL LEARNING IMAGE RESTORATION VIA EFFICIENT GAUSSIAN MIXTURE MODEL LEARNING Jianzhou Feng Li Song Xiaog Huo Xiaokang Yang Wenjun Zhang Shanghai Digital Media Processing Transmission Key Lab, Shanghai Jiaotong University

More information

IMAGE SUPER-RESOLUTION BASED ON DICTIONARY LEARNING AND ANCHORED NEIGHBORHOOD REGRESSION WITH MUTUAL INCOHERENCE

IMAGE SUPER-RESOLUTION BASED ON DICTIONARY LEARNING AND ANCHORED NEIGHBORHOOD REGRESSION WITH MUTUAL INCOHERENCE IMAGE SUPER-RESOLUTION BASED ON DICTIONARY LEARNING AND ANCHORED NEIGHBORHOOD REGRESSION WITH MUTUAL INCOHERENCE Yulun Zhang 1, Kaiyu Gu 2, Yongbing Zhang 1, Jian Zhang 3, and Qionghai Dai 1,4 1 Shenzhen

More information

IMPROVED MOTION-BASED LOCALIZED SUPER RESOLUTION TECHNIQUE USING DISCRETE WAVELET TRANSFORM FOR LOW RESOLUTION VIDEO ENHANCEMENT

IMPROVED MOTION-BASED LOCALIZED SUPER RESOLUTION TECHNIQUE USING DISCRETE WAVELET TRANSFORM FOR LOW RESOLUTION VIDEO ENHANCEMENT 17th European Signal Processing Conference (EUSIPCO 009) Glasgow, Scotland, August 4-8, 009 IMPROVED MOTION-BASED LOCALIZED SUPER RESOLUTION TECHNIQUE USING DISCRETE WAVELET TRANSFORM FOR LOW RESOLUTION

More information

Edge-directed Image Interpolation Using Color Gradient Information

Edge-directed Image Interpolation Using Color Gradient Information Edge-directed Image Interpolation Using Color Gradient Information Andrey Krylov and Andrey Nasonov Laboratory of Mathematical Methods of Image Processing, Faculty of Computational Mathematics and Cybernetics,

More information

NTHU Rain Removal Project

NTHU Rain Removal Project People NTHU Rain Removal Project Networked Video Lab, National Tsing Hua University, Hsinchu, Taiwan Li-Wei Kang, Institute of Information Science, Academia Sinica, Taipei, Taiwan Chia-Wen Lin *, Department

More information

Edge-Directed Image Interpolation Using Color Gradient Information

Edge-Directed Image Interpolation Using Color Gradient Information Edge-Directed Image Interpolation Using Color Gradient Information Andrey Krylov and Andrey Nasonov Laboratory of Mathematical Methods of Image Processing, Faculty of Computational Mathematics and Cybernetics,

More information

Structured Face Hallucination

Structured Face Hallucination 2013 IEEE Conference on Computer Vision and Pattern Recognition Structured Face Hallucination Chih-Yuan Yang Sifei Liu Ming-Hsuan Yang Electrical Engineering and Computer Science University of California

More information

Department of Electronics and Communication KMP College of Engineering, Perumbavoor, Kerala, India 1 2

Department of Electronics and Communication KMP College of Engineering, Perumbavoor, Kerala, India 1 2 Vol.3, Issue 3, 2015, Page.1115-1021 Effect of Anti-Forensics and Dic.TV Method for Reducing Artifact in JPEG Decompression 1 Deepthy Mohan, 2 Sreejith.H 1 PG Scholar, 2 Assistant Professor Department

More information

Sparse Representation Based Super-Resolution Algorithm using Wavelet Domain Interpolation and Nonlocal Means

Sparse Representation Based Super-Resolution Algorithm using Wavelet Domain Interpolation and Nonlocal Means TELKOMNIKA Indonesian Journal of Electrical Engineering Vol. 16, No. 2, November 2015, pp. 296 ~ 302 DOI: 10.11591/telkomnika.v16i2.8816 296 Sparse Representation Based Super-Resolution Algorithm using

More information

Image Super-Resolution Reconstruction Based On L 1/2 Sparsity

Image Super-Resolution Reconstruction Based On L 1/2 Sparsity Buletin Teknik Elektro dan Informatika (Bulletin of Electrical Engineering and Informatics) Vol. 3, No. 3, September 4, pp. 55~6 ISSN: 89-39 55 Image Super-Resolution Reconstruction Based On L / Sparsity

More information

Low Light Image Enhancement via Sparse Representations

Low Light Image Enhancement via Sparse Representations Low Light Image Enhancement via Sparse Representations Konstantina Fotiadou, Grigorios Tsagkatakis, and Panagiotis Tsakalides Institute of Computer Science, Foundation for Research and Technology - Hellas

More information

A Single Image Compression Framework Combined with Sparse Representation-Based Super- Resolution

A Single Image Compression Framework Combined with Sparse Representation-Based Super- Resolution International Conference on Electronic Science and Automation Control (ESAC 2015) A Single Compression Framework Combined with Sparse RepresentationBased Super Resolution He Xiaohai, He Jingbo, Huang Jianqiu

More information

Denoising an Image by Denoising its Components in a Moving Frame

Denoising an Image by Denoising its Components in a Moving Frame Denoising an Image by Denoising its Components in a Moving Frame Gabriela Ghimpețeanu 1, Thomas Batard 1, Marcelo Bertalmío 1, and Stacey Levine 2 1 Universitat Pompeu Fabra, Spain 2 Duquesne University,

More information

MULTI-POSE FACE HALLUCINATION VIA NEIGHBOR EMBEDDING FOR FACIAL COMPONENTS. Yanghao Li, Jiaying Liu, Wenhan Yang, Zongming Guo

MULTI-POSE FACE HALLUCINATION VIA NEIGHBOR EMBEDDING FOR FACIAL COMPONENTS. Yanghao Li, Jiaying Liu, Wenhan Yang, Zongming Guo MULTI-POSE FACE HALLUCINATION VIA NEIGHBOR EMBEDDING FOR FACIAL COMPONENTS Yanghao Li, Jiaying Liu, Wenhan Yang, Zongg Guo Institute of Computer Science and Technology, Peking University, Beijing, P.R.China,

More information

One Network to Solve Them All Solving Linear Inverse Problems using Deep Projection Models

One Network to Solve Them All Solving Linear Inverse Problems using Deep Projection Models One Network to Solve Them All Solving Linear Inverse Problems using Deep Projection Models [Supplemental Materials] 1. Network Architecture b ref b ref +1 We now describe the architecture of the networks

More information

Self-Learning of Edge-Preserving Single Image Super-Resolution via Contourlet Transform

Self-Learning of Edge-Preserving Single Image Super-Resolution via Contourlet Transform Self-Learning of Edge-Preserving Single Image Super-Resolution via Contourlet Transform Min-Chun Yang, De-An Huang, Chih-Yun Tsai, and Yu-Chiang Frank Wang Dept. Computer Science and Information Engineering,

More information

An Optimized Pixel-Wise Weighting Approach For Patch-Based Image Denoising

An Optimized Pixel-Wise Weighting Approach For Patch-Based Image Denoising An Optimized Pixel-Wise Weighting Approach For Patch-Based Image Denoising Dr. B. R.VIKRAM M.E.,Ph.D.,MIEEE.,LMISTE, Principal of Vijay Rural Engineering College, NIZAMABAD ( Dt.) G. Chaitanya M.Tech,

More information

An Improved Approach For Mixed Noise Removal In Color Images

An Improved Approach For Mixed Noise Removal In Color Images An Improved Approach For Mixed Noise Removal In Color Images Ancy Mariam Thomas 1, Dr. Deepa J 2, Rijo Sam 3 1P.G. student, College of Engineering, Chengannur, Kerala, India. 2Associate Professor, Electronics

More information

Image Interpolation Using Multiscale Geometric Representations

Image Interpolation Using Multiscale Geometric Representations Image Interpolation Using Multiscale Geometric Representations Nickolaus Mueller, Yue Lu and Minh N. Do Department of Electrical and Computer Engineering University of Illinois at Urbana-Champaign ABSTRACT

More information

Image Resizing Based on Gradient Vector Flow Analysis

Image Resizing Based on Gradient Vector Flow Analysis Image Resizing Based on Gradient Vector Flow Analysis Sebastiano Battiato battiato@dmi.unict.it Giovanni Puglisi puglisi@dmi.unict.it Giovanni Maria Farinella gfarinellao@dmi.unict.it Daniele Ravì rav@dmi.unict.it

More information

Robust Single Image Super-resolution based on Gradient Enhancement

Robust Single Image Super-resolution based on Gradient Enhancement Robust Single Image Super-resolution based on Gradient Enhancement Licheng Yu, Hongteng Xu, Yi Xu and Xiaokang Yang Department of Electronic Engineering, Shanghai Jiaotong University, Shanghai 200240,

More information

A A A. Fig.1 image patch. Then the edge gradient magnitude is . (1)

A A A. Fig.1 image patch. Then the edge gradient magnitude is . (1) International Conference on Information Science and Computer Applications (ISCA 013) Two-Dimensional Barcode Image Super-Resolution Reconstruction Via Sparse Representation Gaosheng Yang 1,Ningzhong Liu

More information

Exploiting Self-Similarities for Single Frame Super-Resolution

Exploiting Self-Similarities for Single Frame Super-Resolution Exploiting Self-Similarities for Single Frame Super-Resolution Chih-Yuan Yang Jia-Bin Huang Ming-Hsuan Yang Electrical Engineering and Computer Science University of California at Merced Merced, CA 95343,

More information

A Self-Learning Optimization Approach to Single Image Super-Resolution using Kernel ridge regression model

A Self-Learning Optimization Approach to Single Image Super-Resolution using Kernel ridge regression model A Self-Learning Optimization Approach to Single Image Super-Resolution using Kernel ridge regression model Ms. Dharani S 1 PG Student (CSE), Sri Krishna College of Engineering and Technology, Anna University,

More information

Image Quality Assessment Techniques: An Overview

Image Quality Assessment Techniques: An Overview Image Quality Assessment Techniques: An Overview Shruti Sonawane A. M. Deshpande Department of E&TC Department of E&TC TSSM s BSCOER, Pune, TSSM s BSCOER, Pune, Pune University, Maharashtra, India Pune

More information

arxiv: v1 [cs.cv] 3 Jan 2017

arxiv: v1 [cs.cv] 3 Jan 2017 Learning a Mixture of Deep Networks for Single Image Super-Resolution Ding Liu, Zhaowen Wang, Nasser Nasrabadi, and Thomas Huang arxiv:1701.00823v1 [cs.cv] 3 Jan 2017 Beckman Institute, University of Illinois

More information

Boosting face recognition via neural Super-Resolution

Boosting face recognition via neural Super-Resolution Boosting face recognition via neural Super-Resolution Guillaume Berger, Cle ment Peyrard and Moez Baccouche Orange Labs - 4 rue du Clos Courtel, 35510 Cesson-Se vigne - France Abstract. We propose a two-step

More information

Single Image Super-Resolution via Iterative Collaborative Representation

Single Image Super-Resolution via Iterative Collaborative Representation Single Image Super-Resolution via Iterative Collaborative Representation Yulun Zhang 1(B), Yongbing Zhang 1, Jian Zhang 2, aoqian Wang 1, and Qionghai Dai 1,3 1 Graduate School at Shenzhen, Tsinghua University,

More information

Image Inpainting Using Sparsity of the Transform Domain

Image Inpainting Using Sparsity of the Transform Domain Image Inpainting Using Sparsity of the Transform Domain H. Hosseini*, N.B. Marvasti, Student Member, IEEE, F. Marvasti, Senior Member, IEEE Advanced Communication Research Institute (ACRI) Department of

More information

Centralized Sparse Representation for Image Restoration

Centralized Sparse Representation for Image Restoration Centralized Sparse Representation for Image Restoration Weisheng Dong Sch. of Elec. Engineering Xidian University, China wsdong@mail.xidian.edu.cn Lei Zhang Dept. of Computing The Hong Kong Polytechnic

More information

IMPLEMENTATION OF THE CONTRAST ENHANCEMENT AND WEIGHTED GUIDED IMAGE FILTERING ALGORITHM FOR EDGE PRESERVATION FOR BETTER PERCEPTION

IMPLEMENTATION OF THE CONTRAST ENHANCEMENT AND WEIGHTED GUIDED IMAGE FILTERING ALGORITHM FOR EDGE PRESERVATION FOR BETTER PERCEPTION IMPLEMENTATION OF THE CONTRAST ENHANCEMENT AND WEIGHTED GUIDED IMAGE FILTERING ALGORITHM FOR EDGE PRESERVATION FOR BETTER PERCEPTION Chiruvella Suresh Assistant professor, Department of Electronics & Communication

More information

SUPPLEMENTARY MATERIAL

SUPPLEMENTARY MATERIAL SUPPLEMENTARY MATERIAL Zhiyuan Zha 1,3, Xin Liu 2, Ziheng Zhou 2, Xiaohua Huang 2, Jingang Shi 2, Zhenhong Shang 3, Lan Tang 1, Yechao Bai 1, Qiong Wang 1, Xinggan Zhang 1 1 School of Electronic Science

More information

Single Image Super-resolution using Deformable Patches

Single Image Super-resolution using Deformable Patches Single Image Super-resolution using Deformable Patches Yu Zhu 1, Yanning Zhang 1, Alan L. Yuille 2 1 School of Computer Science, Northwestern Polytechnical University, China 2 Department of Statistics,

More information

COMPRESSED FACE HALLUCINATION. Electrical Engineering and Computer Science University of California, Merced, CA 95344, USA

COMPRESSED FACE HALLUCINATION. Electrical Engineering and Computer Science University of California, Merced, CA 95344, USA COMPRESSED FACE HALLUCNATON Sifei Liu Ming-Hsuan Yang Electrical Engineering and Computer Science University of California, Merced, CA 95344, USA ABSTRACT n this paper, we propose an algorithm to hallucinate

More information

Ms.DHARANI SAMPATH Computer Science And Engineering, Sri Krishna College Of Engineering & Technology Coimbatore, India

Ms.DHARANI SAMPATH Computer Science And Engineering, Sri Krishna College Of Engineering & Technology Coimbatore, India Improving Super Resolution of Image by Multiple Kernel Learning Ms.DHARANI SAMPATH Computer Science And Engineering, Sri Krishna College Of Engineering & Technology Coimbatore, India dharanis012@gmail.com

More information

arxiv: v1 [cs.cv] 8 Feb 2018

arxiv: v1 [cs.cv] 8 Feb 2018 DEEP IMAGE SUPER RESOLUTION VIA NATURAL IMAGE PRIORS Hojjat S. Mousavi, Tiantong Guo, Vishal Monga Dept. of Electrical Engineering, The Pennsylvania State University arxiv:802.0272v [cs.cv] 8 Feb 208 ABSTRACT

More information

Image Restoration Using DNN

Image Restoration Using DNN Image Restoration Using DNN Hila Levi & Eran Amar Images were taken from: http://people.tuebingen.mpg.de/burger/neural_denoising/ Agenda Domain Expertise vs. End-to-End optimization Image Denoising and

More information

Deblurring Text Images via L 0 -Regularized Intensity and Gradient Prior

Deblurring Text Images via L 0 -Regularized Intensity and Gradient Prior Deblurring Text Images via L -Regularized Intensity and Gradient Prior Jinshan Pan, Zhe Hu, Zhixun Su, Ming-Hsuan Yang School of Mathematical Sciences, Dalian University of Technology Electrical Engineering

More information

Anchored Neighborhood Regression for Fast Example-Based Super-Resolution

Anchored Neighborhood Regression for Fast Example-Based Super-Resolution Anchored Neighborhood Regression for Fast Example-Based Super-Resolution Radu Timofte 1,2, Vincent De Smet 1, and Luc Van Gool 1,2 1 KU Leuven, ESAT-PSI / iminds, VISICS 2 ETH Zurich, D-ITET, Computer

More information

Structure-adaptive Image Denoising with 3D Collaborative Filtering

Structure-adaptive Image Denoising with 3D Collaborative Filtering , pp.42-47 http://dx.doi.org/10.14257/astl.2015.80.09 Structure-adaptive Image Denoising with 3D Collaborative Filtering Xuemei Wang 1, Dengyin Zhang 2, Min Zhu 2,3, Yingtian Ji 2, Jin Wang 4 1 College

More information

An improved non-blind image deblurring methods based on FoEs

An improved non-blind image deblurring methods based on FoEs An improved non-blind image deblurring methods based on FoEs Qidan Zhu, Lei Sun College of Automation, Harbin Engineering University, Harbin, 5000, China ABSTRACT Traditional non-blind image deblurring

More information

A DEEP DICTIONARY MODEL FOR IMAGE SUPER-RESOLUTION. Jun-Jie Huang and Pier Luigi Dragotti

A DEEP DICTIONARY MODEL FOR IMAGE SUPER-RESOLUTION. Jun-Jie Huang and Pier Luigi Dragotti A DEEP DICTIONARY MODEL FOR IMAGE SUPER-RESOLUTION Jun-Jie Huang and Pier Luigi Dragotti Communications and Signal Processing Group CSP), Imperial College London, UK ABSTRACT Inspired by the recent success

More information

An Approach for Reduction of Rain Streaks from a Single Image

An Approach for Reduction of Rain Streaks from a Single Image An Approach for Reduction of Rain Streaks from a Single Image Vijayakumar Majjagi 1, Netravati U M 2 1 4 th Semester, M. Tech, Digital Electronics, Department of Electronics and Communication G M Institute

More information

Optimizing the Deblocking Algorithm for. H.264 Decoder Implementation

Optimizing the Deblocking Algorithm for. H.264 Decoder Implementation Optimizing the Deblocking Algorithm for H.264 Decoder Implementation Ken Kin-Hung Lam Abstract In the emerging H.264 video coding standard, a deblocking/loop filter is required for improving the visual

More information

Single-image super-resolution in RGB space via group sparse representation

Single-image super-resolution in RGB space via group sparse representation Published in IET Image Processing Received on 10th April 2014 Revised on 22nd September 2014 Accepted on 15th October 2014 ISSN 1751-9659 Single-image super-resolution in RGB space via group sparse representation

More information

Interpolation Based Image Super Resolution by Support-Vector-Regression

Interpolation Based Image Super Resolution by Support-Vector-Regression Interpolation Based Image Super Resolution by Support-Vector-Regression Sowmya. M 1, Anand M.J 2 1 Final Year M.Tech Student, Department Of Electronics And Communication, PES College Of Engineering, Mandya,

More information

Example-Based Image Super-Resolution Techniques

Example-Based Image Super-Resolution Techniques Example-Based Image Super-Resolution Techniques Mark Sabini msabini & Gili Rusak gili December 17, 2016 1 Introduction With the current surge in popularity of imagebased applications, improving content

More information

Fast single image super-resolution based on sigmoid transformation

Fast single image super-resolution based on sigmoid transformation > REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 1 Fast single image super-resolution based on sigmoid transformation Longguang Wang, Zaiping Lin, Jinyan Gao, Xinpu

More information

IEEE TRANSACTIONS ON IMAGE PROCESSING, VOL. 20, NO. 6, JUNE

IEEE TRANSACTIONS ON IMAGE PROCESSING, VOL. 20, NO. 6, JUNE IEEE TRANSACTIONS ON IMAGE PROCESSING, VOL. 20, NO. 6, JUNE 2011 1529 Gradient Profile Prior and Its Applications in Image Super-Resolution and Enhancement Jian Sun, Member, IEEE, Jian Sun, Zongben Xu,

More information

ICA mixture models for image processing

ICA mixture models for image processing I999 6th Joint Sy~nposiurn orz Neural Computation Proceedings ICA mixture models for image processing Te-Won Lee Michael S. Lewicki The Salk Institute, CNL Carnegie Mellon University, CS & CNBC 10010 N.

More information

Comparative Analysis of Edge Based Single Image Superresolution

Comparative Analysis of Edge Based Single Image Superresolution Comparative Analysis of Edge Based Single Image Superresolution Sonali Shejwal 1, Prof. A. M. Deshpande 2 1,2 Department of E&Tc, TSSM s BSCOER, Narhe, University of Pune, India. ABSTRACT: Super-resolution

More information

Learning how to combine internal and external denoising methods

Learning how to combine internal and external denoising methods Learning how to combine internal and external denoising methods Harold Christopher Burger, Christian Schuler, and Stefan Harmeling Max Planck Institute for Intelligent Systems, Tübingen, Germany Abstract.

More information

Non-local Means for Stereo Image Denoising Using Structural Similarity

Non-local Means for Stereo Image Denoising Using Structural Similarity Non-local Means for Stereo Image Denoising Using Structural Similarity Monagi H. Alkinani and Mahmoud R. El-Sakka (B) Computer Science Department, University of Western Ontario, London, ON N6A 5B7, Canada

More information

Optimal Denoising of Natural Images and their Multiscale Geometry and Density

Optimal Denoising of Natural Images and their Multiscale Geometry and Density Optimal Denoising of Natural Images and their Multiscale Geometry and Density Department of Computer Science and Applied Mathematics Weizmann Institute of Science, Israel. Joint work with Anat Levin (WIS),

More information

SSIM Image Quality Metric for Denoised Images

SSIM Image Quality Metric for Denoised Images SSIM Image Quality Metric for Denoised Images PETER NDAJAH, HISAKAZU KIKUCHI, MASAHIRO YUKAWA, HIDENORI WATANABE and SHOGO MURAMATSU Department of Electrical and Electronics Engineering, Niigata University,

More information

Image Super-Resolution via Sparse Representation

Image Super-Resolution via Sparse Representation Image Super-Resolution via Sparse Representation Jianchao Yang, John Wright, Thomas Huang and Yi Ma accepted by IEEE Trans. on Image Processing 2010 Presented by known 2010/4/20 1 Super-Resolution Techniques

More information

Virtual Training Samples and CRC based Test Sample Reconstruction and Face Recognition Experiments Wei HUANG and Li-ming MIAO

Virtual Training Samples and CRC based Test Sample Reconstruction and Face Recognition Experiments Wei HUANG and Li-ming MIAO 7 nd International Conference on Computational Modeling, Simulation and Applied Mathematics (CMSAM 7) ISBN: 978--6595-499-8 Virtual raining Samples and CRC based est Sample Reconstruction and Face Recognition

More information

A Novel Image Super-resolution Reconstruction Algorithm based on Modified Sparse Representation

A Novel Image Super-resolution Reconstruction Algorithm based on Modified Sparse Representation , pp.162-167 http://dx.doi.org/10.14257/astl.2016.138.33 A Novel Image Super-resolution Reconstruction Algorithm based on Modified Sparse Representation Liqiang Hu, Chaofeng He Shijiazhuang Tiedao University,

More information

A Novel Multi-Frame Color Images Super-Resolution Framework based on Deep Convolutional Neural Network. Zhe Li, Shu Li, Jianmin Wang and Hongyang Wang

A Novel Multi-Frame Color Images Super-Resolution Framework based on Deep Convolutional Neural Network. Zhe Li, Shu Li, Jianmin Wang and Hongyang Wang 5th International Conference on Measurement, Instrumentation and Automation (ICMIA 2016) A Novel Multi-Frame Color Images Super-Resolution Framewor based on Deep Convolutional Neural Networ Zhe Li, Shu

More information

Blind Deblurring using Internal Patch Recurrence. Tomer Michaeli & Michal Irani Weizmann Institute

Blind Deblurring using Internal Patch Recurrence. Tomer Michaeli & Michal Irani Weizmann Institute Blind Deblurring using Internal Patch Recurrence Tomer Michaeli & Michal Irani Weizmann Institute Scale Invariance in Natural Images Small patterns recur at different scales Scale Invariance in Natural

More information

Influence of Training Set and Iterative Back Projection on Example-based Super-resolution

Influence of Training Set and Iterative Back Projection on Example-based Super-resolution Influence of Training Set and Iterative Back Projection on Example-based Super-resolution Saeid Fazli Research Institute of Modern Biological Techniques University of zanjan Zanjan, Iran Abstract Example-based

More information

Deep Back-Projection Networks For Super-Resolution Supplementary Material

Deep Back-Projection Networks For Super-Resolution Supplementary Material Deep Back-Projection Networks For Super-Resolution Supplementary Material Muhammad Haris 1, Greg Shakhnarovich 2, and Norimichi Ukita 1, 1 Toyota Technological Institute, Japan 2 Toyota Technological Institute

More information

IMAGE SUPER RESOLUTION USING NON SUB-SAMPLE CONTOURLET TRANSFORM WITH LOCAL TERNARY PATTERN

IMAGE SUPER RESOLUTION USING NON SUB-SAMPLE CONTOURLET TRANSFORM WITH LOCAL TERNARY PATTERN IMAGE SUPER RESOLUTION USING NON SUB-SAMPLE CONTOURLET TRANSFORM WITH LOCAL TERNARY PATTERN Pikin S. Patel 1, Parul V. Pithadia 2, Manoj parmar 3 PG. Student, EC Dept., Dr. S & S S Ghandhy Govt. Engg.

More information

Bayesian region selection for adaptive dictionary-based Super-Resolution

Bayesian region selection for adaptive dictionary-based Super-Resolution PÉREZ-PELLITERO ET AL.: BAYESIAN REGION SELECTION FOR SUPER-RESOLUTION 1 Bayesian region selection for adaptive dictionary-based Super-Resolution Eduardo Pérez-Pellitero 1, 2 eduardo.perezpellitero@technicolor.com

More information

Recovering Realistic Texture in Image Super-resolution by Deep Spatial Feature Transform. Xintao Wang Ke Yu Chao Dong Chen Change Loy

Recovering Realistic Texture in Image Super-resolution by Deep Spatial Feature Transform. Xintao Wang Ke Yu Chao Dong Chen Change Loy Recovering Realistic Texture in Image Super-resolution by Deep Spatial Feature Transform Xintao Wang Ke Yu Chao Dong Chen Change Loy Problem enlarge 4 times Low-resolution image High-resolution image Previous

More information

A Sparse Coding based Approach for the Resolution Enhancement and Restoration of Printed and Handwritten Textual Images

A Sparse Coding based Approach for the Resolution Enhancement and Restoration of Printed and Handwritten Textual Images A Sparse Coding based Approach for the Resolution Enhancement and Restoration of Printed and Handwritten Textual Images Rim Walha, Fadoua Drira and Adel M. Alimi University of Sfax, ENI-Sfax, REGIM Sfax,

More information

Blind Image Deblurring Using Dark Channel Prior

Blind Image Deblurring Using Dark Channel Prior Blind Image Deblurring Using Dark Channel Prior Jinshan Pan 1,2,3, Deqing Sun 2,4, Hanspeter Pfister 2, and Ming-Hsuan Yang 3 1 Dalian University of Technology 2 Harvard University 3 UC Merced 4 NVIDIA

More information

3194 IEEE TRANSACTIONS ON IMAGE PROCESSING, VOL. 25, NO. 7, JULY 2016

3194 IEEE TRANSACTIONS ON IMAGE PROCESSING, VOL. 25, NO. 7, JULY 2016 3194 IEEE TRANSACTIONS ON IMAGE PROCESSING, VOL. 25, NO. 7, JULY 2016 Robust Single Image Super-Resolution via Deep Networks With Sparse Prior Ding Liu, Student Member, IEEE, Zhaowen Wang, Member, IEEE,

More information

Inverse Problems and Machine Learning

Inverse Problems and Machine Learning Inverse Problems and Machine Learning Julian Wörmann Research Group for Geometric Optimization and Machine Learning (GOL) 1 What are inverse problems? 2 Inverse Problems cause/ excitation 3 Inverse Problems

More information

Image Processing Via Pixel Permutations

Image Processing Via Pixel Permutations Image Processing Via Pixel Permutations Michael Elad The Computer Science Department The Technion Israel Institute of technology Haifa 32000, Israel Joint work with Idan Ram Israel Cohen The Electrical

More information

GRID WARPING IN TOTAL VARIATION IMAGE ENHANCEMENT METHODS. Andrey Nasonov, and Andrey Krylov

GRID WARPING IN TOTAL VARIATION IMAGE ENHANCEMENT METHODS. Andrey Nasonov, and Andrey Krylov GRID WARPING IN TOTAL VARIATION IMAGE ENHANCEMENT METHODS Andrey Nasonov, and Andrey Krylov Lomonosov Moscow State University, Moscow, Department of Computational Mathematics and Cybernetics, e-mail: nasonov@cs.msu.ru,

More information

On Single Image Scale-Up using Sparse-Representation

On Single Image Scale-Up using Sparse-Representation On Single Image Scale-Up using Sparse-Representation Roman Zeyde, Matan Protter and Michael Elad The Computer Science Department Technion Israel Institute of Technology Haifa 32000, Israel {romanz,matanpr,elad}@cs.technion.ac.il

More information

A Comparison of Still-Image Compression Standards Using Different Image Quality Metrics and Proposed Methods for Improving Lossy Image Quality

A Comparison of Still-Image Compression Standards Using Different Image Quality Metrics and Proposed Methods for Improving Lossy Image Quality A Comparison of Still-Image Compression Standards Using Different Image Quality Metrics and Proposed Methods for Improving Lossy Image Quality Multidimensional DSP Literature Survey Eric Heinen 3/21/08

More information

Fast and Effective Interpolation Using Median Filter

Fast and Effective Interpolation Using Median Filter Fast and Effective Interpolation Using Median Filter Jian Zhang 1, *, Siwei Ma 2, Yongbing Zhang 1, and Debin Zhao 1 1 Department of Computer Science, Harbin Institute of Technology, Harbin 150001, P.R.

More information

A Comparative Analysis of Noise Reduction Filters in Images Mandeep kaur 1, Deepinder kaur 2

A Comparative Analysis of Noise Reduction Filters in Images Mandeep kaur 1, Deepinder kaur 2 A Comparative Analysis of Noise Reduction Filters in Images Mandeep kaur 1, Deepinder kaur 2 1 Research Scholar, Dept. Of Computer Science & Engineering, CT Institute of Technology & Research, Jalandhar,

More information

A New Orthogonalization of Locality Preserving Projection and Applications

A New Orthogonalization of Locality Preserving Projection and Applications A New Orthogonalization of Locality Preserving Projection and Applications Gitam Shikkenawis 1,, Suman K. Mitra, and Ajit Rajwade 2 1 Dhirubhai Ambani Institute of Information and Communication Technology,

More information

Bilevel Sparse Coding

Bilevel Sparse Coding Adobe Research 345 Park Ave, San Jose, CA Mar 15, 2013 Outline 1 2 The learning model The learning algorithm 3 4 Sparse Modeling Many types of sensory data, e.g., images and audio, are in high-dimensional

More information

FAST SINGLE-IMAGE SUPER-RESOLUTION WITH FILTER SELECTION. Image Processing Lab Technicolor R&I Hannover

FAST SINGLE-IMAGE SUPER-RESOLUTION WITH FILTER SELECTION. Image Processing Lab Technicolor R&I Hannover FAST SINGLE-IMAGE SUPER-RESOLUTION WITH FILTER SELECTION Jordi Salvador Eduardo Pérez-Pellitero Axel Kochale Image Processing Lab Technicolor R&I Hannover ABSTRACT This paper presents a new method for

More information

Bidirectional Recurrent Convolutional Networks for Video Super-Resolution

Bidirectional Recurrent Convolutional Networks for Video Super-Resolution Bidirectional Recurrent Convolutional Networks for Video Super-Resolution Qi Zhang & Yan Huang Center for Research on Intelligent Perception and Computing (CRIPAC) National Laboratory of Pattern Recognition

More information

Image Super-Resolution Using Compressed Sensing Based on Learning Sub Dictionary

Image Super-Resolution Using Compressed Sensing Based on Learning Sub Dictionary Image Super-Resolution Using Compressed Sensing Based on Learning Sub Dictionary 1 MUHAMMAD SAMEER SHEIKH, 1 QUNSHENG CAO, 2 CAIYUN WANG 1 College of Electronics and Information Engineering, 2 College

More information

Enhancing DubaiSat-1 Satellite Imagery Using a Single Image Super-Resolution

Enhancing DubaiSat-1 Satellite Imagery Using a Single Image Super-Resolution Enhancing DubaiSat-1 Satellite Imagery Using a Single Image Super-Resolution Saeed AL-Mansoori 1 and Alavi Kunhu 2 1 Associate Image Processing Engineer, SIPAD Image Enhancement Section Emirates Institution

More information

Texture Sensitive Image Inpainting after Object Morphing

Texture Sensitive Image Inpainting after Object Morphing Texture Sensitive Image Inpainting after Object Morphing Yin Chieh Liu and Yi-Leh Wu Department of Computer Science and Information Engineering National Taiwan University of Science and Technology, Taiwan

More information

Face Hallucination Based on Eigentransformation Learning

Face Hallucination Based on Eigentransformation Learning Advanced Science and Technology etters, pp.32-37 http://dx.doi.org/10.14257/astl.2016. Face allucination Based on Eigentransformation earning Guohua Zou School of software, East China University of Technology,

More information

Generalized Tree-Based Wavelet Transform and Applications to Patch-Based Image Processing

Generalized Tree-Based Wavelet Transform and Applications to Patch-Based Image Processing Generalized Tree-Based Wavelet Transform and * Michael Elad The Computer Science Department The Technion Israel Institute of technology Haifa 32000, Israel *Joint work with A Seminar in the Hebrew University

More information

PRINCIPAL COMPONENT ANALYSIS IMAGE DENOISING USING LOCAL PIXEL GROUPING

PRINCIPAL COMPONENT ANALYSIS IMAGE DENOISING USING LOCAL PIXEL GROUPING PRINCIPAL COMPONENT ANALYSIS IMAGE DENOISING USING LOCAL PIXEL GROUPING Divesh Kumar 1 and Dheeraj Kalra 2 1 Department of Electronics & Communication Engineering, IET, GLA University, Mathura 2 Department

More information

BLIND IMAGE QUALITY ASSESSMENT WITH LOCAL CONTRAST FEATURES

BLIND IMAGE QUALITY ASSESSMENT WITH LOCAL CONTRAST FEATURES BLIND IMAGE QUALITY ASSESSMENT WITH LOCAL CONTRAST FEATURES Ganta Kasi Vaibhav, PG Scholar, Department of Electronics and Communication Engineering, University College of Engineering Vizianagaram,JNTUK.

More information

REJECTION-BASED CLASSIFICATION FOR ACTION RECOGNITION USING A SPATIO-TEMPORAL DICTIONARY. Stefen Chan Wai Tim, Michele Rombaut, Denis Pellerin

REJECTION-BASED CLASSIFICATION FOR ACTION RECOGNITION USING A SPATIO-TEMPORAL DICTIONARY. Stefen Chan Wai Tim, Michele Rombaut, Denis Pellerin REJECTION-BASED CLASSIFICATION FOR ACTION RECOGNITION USING A SPATIO-TEMPORAL DICTIONARY Stefen Chan Wai Tim, Michele Rombaut, Denis Pellerin Univ. Grenoble Alpes, GIPSA-Lab, F-38000 Grenoble, France ABSTRACT

More information

A Novel Approach for Deblocking JPEG Images

A Novel Approach for Deblocking JPEG Images A Novel Approach for Deblocking JPEG Images Multidimensional DSP Final Report Eric Heinen 5/9/08 Abstract This paper presents a novel approach for deblocking JPEG images. First, original-image pixels are

More information

Locally Adaptive Learning for Translation-Variant MRF Image Priors

Locally Adaptive Learning for Translation-Variant MRF Image Priors Locally Adaptive Learning for Translation-Variant MRF Image Priors Masayuki Tanaka and Masatoshi Okutomi Tokyo Institute of Technology 2-12-1 O-okayama, Meguro-ku, Tokyo, JAPAN mtanaka@ok.ctrl.titech.ac.p,

More information