Multiple Dictionaries Research Articles

A sparse representation-based image resolution improvement method by processing multiple dictionary pairs with latent Dirichlet allocation model for street view images

Street view applications are widely used in many situations. However, the resolution of the street view image is not high enough. Users always desire high resolution street view images. Image resolution improvement methods can effectively generate a high resolution street view image from a single low resolution street view image. The sparse representation-based image resolution improvement method is a promising way to improve the resolution of an image. However, only one dictionary pair, which fails to represent the diverse structures in images, is used in conventional sparse representation-based methods This may lead to poor performances in many circumstances. In this paper, we propose a new sparse representation-based method with multiple dictionary pairs. To capture the various structures at the semantic level, our method adopts latent Dirichlet allocation model to divide the patches into clusters. Then we learn a dictionary pair for each cluster. Finally, these dictionary pairs are used to reconstruct high resolution images. Experimental results validate that our method is superior over the compared methods in both visual perception and objective quantitation.

Sparse coding of i-vector/JFA latent vector over ensemble dictionaries for language identification systems

Computing sparse representation (SR) over an exemplar dictionary is time consuming and computationally expensive for large dictionary size. This also requires huge memory requirement for saving the dictionary. In order to reduce the latency and to achieve some diversity, ensemble of exemplar dictionary based language identification (LID) system is explored. The full diversity can be obtained if each of the exemplar dictionary contains only one feature vector from each of the language class. To achieve full diversity, a large number of multiple dictionaries are required; thus needs to compute SR for a particular test utterance as many times. The other solution to reduce the latency is to use a learned dictionary. The dictionary may contain unequal number of dictionary atoms and it is not guaranteed that each language class information is present. It totally depends upon the number of data and its variations. Motivated by this, language specific dictionary is learned, and then concatenated to form a single learned dictionary. Furthermore, to overcome the problem of ensemble exemplar dictionary based LID system, we investigated the ensemble of learned-exemplar dictionary based LID system. The proposed approach achieves the same diversity and latency as that of ensemble exemplar dictionary with reduced number of learned dictionaries. The proposed techniques are applied on two spoken utterance representations: the i-vector and the JFA latent vector. The experiments are performed on 2007 NIST LRE, 2009 NIST LRE and AP17-OLR datasets in closed set condition.

Spatio-Temporal Series Remote Sensing Image Prediction Based on Multi-Dictionary Bayesian Fusion

Contradictions in spatial resolution and temporal coverage emerge from earth observation remote sensing images due to limitations in technology and cost. Therefore, how to combine remote sensing images with low spatial yet high temporal resolution as well as those with high spatial yet low temporal resolution to construct images with both high spatial resolution and high temporal coverage has become an important problem called spatio-temporal fusion problem in both research and practice. A Multi-Dictionary Bayesian Spatio-Temporal Reflectance Fusion Model (MDBFM) has been proposed in this paper. First, multiple dictionaries from regions of different classes are trained. Second, a Bayesian framework is constructed to solve the dictionary selection problem. A pixel-dictionary likehood function and a dictionary-dictionary prior function are constructed under the Bayesian framework. Third, remote sensing images before and after the middle moment are combined to predict images at the middle moment. Diverse shapes and textures information is learned from different landscapes in multi-dictionary learning to help dictionaries capture the distinctions between regions. The Bayesian framework makes full use of the priori information while the input image is classified. The experiments with one simulated dataset and two satellite datasets validate that the MDBFM is highly effective in both subjective and objective evaluation indexes. The results of MDBFM show more precise details and have a higher similarity with real images when dealing with both type changes and phenology changes.

Medical image super-resolution by using multi-dictionary and random forest

Smart City has become the direction of the development of city. Telemedicine is an important part of Smart City. Telemedicine always provides clinical health care according to the medical images of the patient. High resolution images are expected for remote diagnosis. Super-resolution technology can improve the resolution of medical images. Recently, sparse coding based super-resolution has attracted more attentions. Sparse coding based super-resolution tries to find the sparse representation of low-resolution (LR) image patches from low resolution dictionary, then reconstructs high-resolution (HR) image patches using sparse representation and HR dictionary. In this paper, we propose a sparse-based scheme for medical image super-resolution. First, we jointly divide the training patches into several clusters. Multiple dictionaries are learned from each cluster to collectively provide the least super-resolution error for the training patches. Second, random forest is trained based on the training patches and their cluster labels. Finally, for an input LR image patches, we use trained random forest to determine which cluster the patch belong to, then use the corresponding dictionary to reconstruct the patch. Thus, all the input LR patches are reconstructed with smallest error. All the reconstructed HR patches are synthesized into a completed HR image. The proposed scheme is applied to test a set of medical images. Experimental results show that both objective evaluation (PSNR) and subjective evaluation (visual effect) are improved when compare to other example-based methods.

Spectral Unmixing With Multiple Dictionaries

Spectral unmixing aims at recovering the spectral signatures of materials, called endmembers, mixed in a hyperspectral or multispectral image, along with their abundances. A typical assumption is that the image contains one pure pixel per endmember, in which case spectral unmixing reduces to identifying these pixels. Many fully automated methods have been proposed in recent years, but little work has been done to allow users to select areas where pure pixels are present manually or using a segmentation algorithm. Additionally, in a non-blind approach, several spectral libraries may be available rather than a single one, with a fixed number (or an upper or lower bound) of endmembers to chose from each. In this paper, we propose a multiple-dictionary constrained low-rank matrix approximation model that address these two problems. We propose an algorithm to compute this model, dubbed M2PALS, and its performance is discussed on both synthetic and real hyperspectral images.

RETRACTED ARTICLE: Single image shadow detection and removal based on feature fusion and multiple dictionary learning

In recent years, the analysis of natural image has made great progress while the image of the intrinsic component analysis can solve many computer vision problems, such as the image shadow detection and removal. This paper presents the novel model, which integrates the feature fusion and the multiple dictionary learning. Traditional model can hardly handle the challenge of reserving the removal accuracy while keeping the low time consuming. Inspire by the compressive sensing theory, traditional single dictionary scenario is extended to the multiple condition. The human visual system is more sensitive to the high frequency part of the image, and the high frequency part expresses most of the semantic information of the image. At the same time, the high frequency characteristic of the high and low resolution image is adopted in the dictionary training, which can effectively recover the loss in the high resolution image with high frequency information. This paper presents the integration of compressive sensing model with feature extraction to construct the two-stage methodology. Therefore, the feature fusion algorithm is applied to the dictionary training procedure to finalize the robust model. Simulation results proves the effectiveness of the model, which outperforms compared with the other state-of-the-art algorithms.

Compact MQDF classifiers using sparse coding for handwritten Chinese character recognition

The modified quadratic discriminant function (MQDF) is an effective classifier for handwritten Chinese character recognition (HCCR). However, it suffers from high memory requirement for the storage of its parameters, which makes it impractical to be embedded in memory limited hand-held devices. In this paper, we explore the applicability of sparse coding to build compact MQDF classifiers. To be specific, we use sparse coding to compact the parameters of MQDF. Two methods of sparse coding, viz., the maximum likelihood-based method and the K-SVD method, are adopted to build two compact MQDF classifiers, namely, MQDF-ML classifier and MQDF-KSVD classifier. Furthermore, we learn multiple dictionaries rather than single dictionary for sparse coding, because the multiple dictionary learning is capable of not only greatly reducing the computational complexity, but also alleviating the degradation of recognition accuracy, compared to the single dictionary learning. Experiments and comparison with the existing method have demonstrated the effectiveness of our proposed method for the issue of unconstrained handwritten Chinese character recognition.

Coupled dictionary learning in wavelet domain for Single-Image Super-Resolution

In this paper a coupled dictionary learning mechanism with mapping function is proposed in the wavelet domain for the task of Single-Image Super-Resolution. Sparsity is used as the invariant feature for achieving super-resolution. Instead of using a single dictionary multiple compact dictionaries are proposed in the wavelet domain. Such dictionaries will exhibit the properties of the wavelet transform such as compactness, directionality and redundancy. Six pairs of dictionaries are designed using a coupled dictionary mechanism with mapping function which helps in strengthening the similarity between the sparse coefficients. Low-resolution image is assumed as the approximation image of the first-level wavelet decomposition. High resolution is achieved by estimating the wavelet sub-bands of this low-resolution image by dictionary learning and sparsity. The proposed algorithm outperforms a well-known spatial domain and wavelet domain algorithm as evaluated on the existing comparative parameters such as structural similarity index measure and peak signal-to-noise ratio.

Multiple dictionary pairs learning and sparse representation-based infrared image super-resolution with improved fuzzy clustering

Exploring sparse representation to enhance the resolution of infrared image has attracted much attention in the last decade. However, conventional sparse representation-based super-resolution aim at learning a universal and efficient dictionary pair for image representation. However, considering that a large number of different structures exist in an image, it is insufficient and unreasonable to present various image structures with only one universal dictionary pair. In this paper, we propose an improved fuzzy clustering and weighted scheme reconstruction framework to solve this problem. Firstly, the training patches are divided into multiple clusters by joint learning multiple dictionary pairs with improved fuzzy clustering method. The goal of joint learning is to learn the multiple dictionary pairs which could collectively represent all the training patches with smallest reconstruction error. So that the learned dictionary pairs are more precise and mutually complementary. Then, high-resolution (HR) patches are estimated according to several most accurate dictionary pairs. Finally, these estimated HR patches are integrated together to generate a final HR patch by a weighted scheme. Numerous experiments demonstrate that this framework outperforms some state-of-art super-resolution methods in both quantitatively and perceptually.

Superresolution for UAV Images via Adaptive Multiple Sparse Representation and Its Application to 3-D Reconstruction

We propose a superresolution (SR) algorithm based on adaptive sparse representation via multiple dictionaries for images taken by unmanned aerial vehicles (UAVs). The SR attainable through the proposed algorithm can increase the precision of 3-D reconstruction from UAV images, enabling the production of high-resolution images for constructing high-frequency time series and for high-precision digital mapping in agriculture. The basic idea of the proposed method is to use a field server or ground-based camera to take training images and then construct multiple pairs of dictionaries based on selective sparse representations to reduce instability during the sparse coding process. The dictionaries are classified on the basis of the edge orientation into five clusters: 0, 45, 90, 135, and nondirection. The proposed method is expected to reduce blurring, blocking, and ringing artifacts especially in edge areas. We evaluated the proposed and previous methods using peak signal-to-noise ratio, structural similarity, feature similarity, and computation time. Our experimental results indicate that the proposed method clearly outperforms other state-of-the-art algorithms based on qualitative and quantitative analysis. In the end, we demonstrate the effectiveness of our proposed method to increase the precision of 3-D reconstruction from UAV images.

Robust Visual Tracking With Multitask Joint Dictionary Learning

Dictionary learning for sparse representation has been increasingly applied to object tracking, however, the existing methods only utilize one modality of the object to learn a single dictionary. In this paper, we propose a robust tracking method based on multitask joint dictionary learning. Through extracting different features of the target, multiple linear sparse representations are obtained. Each sparse representation can be learned by a corresponding dictionary. Instead of separately learning the multiple dictionaries, we adopt a multitask learning approach to learn the multiple linear sparse representations, which provide additional useful information to the classification problem. Because different tasks may favor different sparse representation coefficients, yet the joint sparsity may enforce the robustness in coefficient estimation. During tracking, a classifier is constructed based on a joint linear representation, and the candidate with the smallest joint decision error is selected to be the tracked object. In addition, reliable tracking results and augmented training samples are accumulated into two sets to update the dictionaries for classification, which helps our tracker adapt to the fast time-varying object appearance. Both qualitative and quantitative evaluations on CVPR2013 visual tracking benchmark demonstrate that our method performs favorably against state-of-the-art trackers.

Multi-sensor image super-resolution with fuzzy cluster by using multi-scale and multi-view sparse coding for infrared image

Super-resolution (SR) methods are effective for generating a high-resolution image from a single low-resolution image. However, four problems are observed in existing SR methods. (1) They cannot reconstruct many details from a low-resolution infrared image because infrared images always lack detailed information. (2) They cannot extract the desired information from images because they do not consider that images naturally come at different scales in many cases. (3) They fail to reveal different physical structures of low-resolution patch because they extract features from a single view. (4) They fail to extract all the different patterns because they use only one dictionary to represent all patterns. To overcome these problems, we propose a novel SR method for infrared images. First, we combine the information of high-resolution visible light images and low-resolution infrared images to improve the resolution of infrared images. Second, we use multiscale patches instead of fixed-size patches to represent infrared images more accurately. Third, we use different feature vectors rather than a single feature to represent infrared images. Finally, we divide training patches into several clusters, and multiple dictionaries are learned for each cluster to provide each patch with a more accurate dictionary. In the proposed method, clustering information for low-resolution patches is learnt by using fuzzy clustering theory. Experiments validate that the proposed method yields better results in terms of quantization and visual perception than the state-of-the-art algorithms.

Informed Group-Sparse Representation for Singing Voice Separation

Singing voice separation attempts to separate the vocal and instrumental parts of a music recording, which is a fundamental problem in music information retrieval. Recent work on singing voice separation has shown that the low-rank representation and informed separation approaches are both able to improve separation quality. However, low-rank optimizations are computationally inefficient due to the use of singular value decompositions. Therefore, in this paper, we propose a new linear-time algorithm called informed group-sparse representation, and use it to separate the vocals from music using pitch annotations as side information. Experimental results on the iKala dataset confirm the efficacy of our approach, suggesting that the music accompaniment follows a group-sparse structure given a pre-trained instrumental dictionary. We also show how our work can be easily extended to accommodate multiple dictionaries using the DSD100 dataset.

Employing structural and statistical information to learn dictionary(s) for single image super-resolution in sparse domain

It has been argued that structural information plays a significant role in the perceptual quality of images, but the importance of statistical information cannot be neglected. In this work, we have proposed an approach, which explores both structural and statistical information of image patches to learn multiple dictionaries for super-resolving an image in sparse domain. Structural information is estimated using dominant edge orientation, and mean value of the intensity levels of an image patch is used to represent statistical information. During reconstruction, a low resolution test patch is inspected for its structural as well as statistical information to choose a suitable dictionary. This helps in preserving the orientation of edge during super-resolution process. Results are further improved by adding an edge (magnitude of edge) preserving constraint, which maintains the edge continuity of super-resolved image with the input low resolution image. Thus, both characteristics of edge, i.e., orientation and magnitude are preserved in our proposed approach. The experimental results demonstrate the usefulness of the proposed approach in comparison to state-of-the-art approaches.

Text-image super-resolution through anchored neighborhood regression with multiple class-specific dictionaries

In the dictionary-based image super-resolution (SR) methods, the resolution of the input image is enhanced using a dictionary of low-resolution (LR) and high-resolution (HR) image patches. Typically, a single dictionary is learned from all the patches in the training set. Then, the input LR patch is super-resolved using its nearest LR patches and their corresponding HR patches in the dictionary. In this paper, we propose a text-image SR method using multiple class-specific dictionaries. Each dictionary is learned from the patches of images of a specific character in the training set. The input LR image is segmented into text lines and characters, and the characters are preliminarily classified. Likewise, overlapping patches are extracted from the input LR image. Then, each patch is super-resolved through the anchored neighborhood regression, using n class-specific dictionaries corresponding to the top-n classification results of the character containing the patch. The final HR image is generated by aggregating all the super-resolved patches. Our method achieves significant improvements in visual image quality and OCR accuracy, compared to the related dictionary-based SR methods. This confirms the effectiveness of applying the preliminary character classification results and multiple class-specific dictionaries in text-image SR.

A phase congruency based patch evaluator for complexity reduction in multi-dictionary based single-image super-resolution

Single-image based super-resolution (SISR) aims to recover a high-resolution (HR) image from one of its degraded low-resolution (LR) images. To improve the quality of reconstructed HR image, many researchers attempt to adopt multiple pairs of dictionaries to sparsely represent the image patches. Conventionally, all the patches with different contents are treated equally, and each patch is coded by multiple pairs of dictionaries, which results in tremendous computational burden in the reconstruction process. In this paper, a phase congruency (PC) based patch evaluator (PCPE) is proposed to divide the LR patches into three categories: significant, less-significant and smooth based on the complexity of the contents. Thus, a flexible multi-dictionary based SISR (MDSISR) framework is proposed, which reconstructs different patches by different approaches. In this framework, multiple dictionaries are only applied to scale up the significant patches to maintain high reconstruction accuracy. Also, two simpler baseline approaches are used to reconstruct the less-significant and smooth patches, respectively. Experimental studies on benchmark database demonstrate that the proposed method can achieve competitive PSNR, SSIM, and FSIM with some state-of-the-art SISR approaches. Besides, it can reduce the computational cost in conventional MDSISR significantly without much degradation in visual and numerical results.

Multitask joint spatial pyramid matching using sparse representation with dynamic coefficients for object recognition

Object recognition is considered a necessary part in many computer vision applications. Recently, sparse coding methods, based on representing a sparse feature from an image, show remarkable results on several object recognition benchmarks, but the precision obtained by these methods is not yet sufficient. Such a problem arises where there are few training images available. As such, using multiple features and multitask dictionaries appears to be crucial to achieving better results. We use multitask joint sparse representation, using dynamic coefficients to connect these sparse features. In other words, we calculate the importance of each feature for each class separately. This causes the features to be used efficiently and appropriately for each class. Thus, we use variance of features and particle swarm optimization algorithms to obtain these dynamic coefficients. Experimental results of our work on Caltech-101 and Caltech-256 databases show more accuracy compared with state-of-the art ones on the same databases.

A Machine Learning Approach for Removal of JPEG Compression Artifacts: A Survey

JPEG is a widely used image compression method. Though it is very efficient, it introduces certain artifacts and quantization noise. This paper is a detailed survey about various existing methods for the reduction of these artifacts. The paper explains each method and their advantages and drawbacks. Some of the methods mentioned are Weiner filtering, Image Optimization, Zero-masking, Local Edge regeneration, Multiple dictionary learning, Hybrid Filtering, Fuzzy filtering, Total Variation Regularization, Offset and Shift Technique, Post-processing et al. Also, a comparative study is made as to which method is suitable for which scenario.

Adaptive Dictionary Reconstruction for Compressed Sensing of ECG Signals.

This paper proposes a novel adaptive dictionary (AD) reconstruction scheme to improve the performance of compressed sensing (CS) with electrocardiogram signals (ECG). The method is based on the use of multiple dictionaries, created using dictionary learning (DL) techniques for CS signal reconstruction. The modified reconstruction framework is a two-stage process that leverages information about the signal from an initial signal reconstruction stage. By identifying whether a QRS complex is present and if so, determining a location estimate of the QRS, the most appropriate dictionary is selected and a second stage more refined signal reconstruction can be obtained. The performance of the proposed algorithm is compared with state-of-the-art CS implementations in the literature, as well as the set partitioning in hierarchical trees (SPIHT) wavelet-based lossy compression algorithm. The results indicate that the proposed reconstruction scheme outperforms all existing CS implementations in terms of signal fidelity at each compression ratio tested. The performance of the proposed approach also compares favorably with SPIHT in terms of signal reconstruction quality. Furthermore, an analysis of the overall power consumption of the proposed ECG compression framework as would be used in a body area network (BAN) demonstrates positive results for the proposed CS approach when compared with existing CS techniques and SPIHT.

Analysis SimCO Algorithms for Sparse Analysis Model Based Dictionary Learning

In this paper, we consider the dictionary learning problem for the sparse analysis model. A novel algorithm is proposed by adapting the simultaneous codeword optimization (SimCO) algorithm, based on the sparse synthesis model, to the sparse analysis model. This algorithm assumes that the analysis dictionary contains unit l2-norm atoms and learns the dictionary by optimization on manifolds. This framework allows multiple dictionary atoms to be updated simultaneously in each iteration. However, similar to several existing analysis dictionary learning algorithms, dictionaries learned by the proposed algorithm may contain similar atoms, leading to a degenerate (coherent) dictionary. To address this problem, we also consider restricting the coherence of the learned dictionary and propose Incoherent Analysis SimCO by introducing an atom decorrelation step following the update of the dictionary. We demonstrate the competitive performance of the proposed algorithms using experiments with synthetic data and image denoising as compared with existing algorithms.