Sparse Feature Representation Research Articles

Learning sparse feature representation for blind quality assessment of night-time images

Capturing Night-Time Images (NTIs) with high-quality is quite challenging for consumer photography and several practical applications. Thus, addressing the quality assessment of night-time images is urgently needed. Since there is no available reference image for such images, Night-Time image Quality Assessment (NTQA) should be done blindly. Although Blind natural Image Quality Assessment (BIQA) has attracted a great deal of attention for a long time, very little work has been done in the field of NTQA. Due to the capturing conditions, NTIs suffer from various complex authentic distortions that make it a challenging field of research. Therefore, previous BIQA methods, do not provide sufficient correlation with subjective scores in the case of NTIs and special methods of NTQA should be developed. In this paper we conduct an unsupervised feature learning method for blind quality assessment of night-time images. The features are the sparse representation over the data-adaptive dictionaries learned on the image exposure and gradient magnitude maps. Having these features, an ensemble regression model trained using least squares gradient boosting scheme predicts high correlated objective scores on the standard datasets.

Research on Computer Artificial Intelligence Algorithm to Optimize Physical Training Intensity and Recovery Cycle Model

This paper established an association model between high-load training and cardiac damage to minimize sports injuries. It is a key condition to promote the development of modern sports. This project intends to use 8 biochemical parameters related to high-load training to build a mathematical model of high-load training-myocardial damage. A mathematical model for high load training and cardiac damage was established. Firstly, based on sparse feature representation method, the biochemical mechanism model of high-intensity training was constructed. The energy supply law of intramuscular energy supply and the dynamic law of glycolysis energy supply during exercise were extracted. Eight biochemical parameters were obtained after high intensity training. To construct an accurate association between high-intensity training-heart damage by measuring the correlation between eight indicators of performance. The validity of the model is verified by simulation.

Feature-oriented unified dictionary learning-based sparse classification for multi-domain fault diagnosis

Aiming at the instability of sparse representation of multi-domain features caused by domain discrepancy and individual differences in feature sparse discriminative ability, the Feature-oriented Unified Dictionary Learning based Sparse Classification (FUDL-SC) is proposed for multi-domain fault diagnosis. Blending sources alignment is designed to unify the transformations of multiple domains at the sample global level to reduce the discrepancy in the data distribution across domains. The FUDL framework is established by learning dictionary atoms specific to each feature of different classes in the unified domain, thus generating a Feature-oriented Unified Dictionary (FUD). Reconstruction Scoring Matrix (RSM) is constructed to measure the sparse discrimination performance of individual FUD. The iterative update of the RSM is incorporated into the FUDL model, thus enhancing FUD's sparsity-preserving and discriminating ability in multiple domains. The efficacy of FUDL-SC is experimentally demonstrated on a multi-condition bearing failure dataset, a multi-device bearing failure dataset, and a small-sample gearbox failure dataset. The comparative studies verify that FUDL-SC can achieve higher recognition accuracy, better stability, and greater robustness than numerous state-of-the-art methods in multi-domain fault diagnosis.

Low-Shot Unsupervised Visual Anomaly Detection via Sparse Feature Representation.

Visual anomaly detection is an essential component in modern industrial manufacturing. Existing studies using notions of pairwise similarity distance between a test feature and nominal features have achieved great breakthroughs. However, the absolute similarity distance lacks certain generalizations, making it challenging to extend the comparison beyond the available samples. This limitation could potentially hamper anomaly detection performance in scenarios with limited samples. This article presents a novel sparse feature representation anomaly detection (SFRAD) framework, which formulates the anomaly detection as a sparse feature representation problem; and notably proposes an anomaly score by orthogonal matching pursuit (ASOMP) as a novel detection metric. Specifically, SFRAD calculates the Gaussian kernel distance between the test feature and its sparse representation in the nominal feature space for anomaly detection. Here, the orthogonal matching pursuit (OMP) algorithm is adopted to achieve the sparse feature representation. Moreover, to construct a low-redundancy memory bank storing the basis features for sparse representation, a novel basis feature sampling (BFS) algorithm is proposed by considering both the maximum coverage and the optimum feature representation simultaneously. As a result, SFRAD incorporates both the advantages of absolute similarity and linear representation; and this enhances the generalization in low-shot scenarios. Extensive experiments on the MVTec anomaly detection (MVTec AD), Kolektor surface-defect dataset (KolektorSDD), Kolektor surface-defect dataset 2 (KolektorSDD2), MVTec logical constraints anomaly detection (MVTec LOCO AD), Visual anomaly (VISA), Modified national institute of standards and technology (MNIST), and CIFAR-10 datasets demonstrate that our proposed SFRAD outperforms the previous methods and achieves state-of-the-art unsupervised anomaly detection performance. Notably, significantly improved outcomes and results have also been achieved on low-shot anomaly detection. Code is available at https://github.com/fanghuisky/SFRAD.

A novel random spectral similar component decomposition method and its application to gear fault diagnosis

Sparse random mode decomposition (SRMD) is a decomposition approach established by combining sparse random feature model with clustering algorithm. It is not subject to the sampling process of signal and can mitigate mode mixing. However, the performance of SRMD is limited by its own hyperparameters, and it is prone to derive inaccurate clustering decomposition results when processing strong noise interference signal. To overcome these defects, a novel method called random spectral similar component decomposition (RSSCD) is proposed. In RSSCD, the time–frequency localized features produced by randomization and sparsification are adopted to represent the input signal. Subsequently, the initial signal components formed by sparse random features are taken as a whole, and the spectral similarity criterion is defined to adaptively form independent random components (RCs), thus improving the accuracy of decomposition. Furthermore, RSSCD is applied to gear fault diagnosis, and a fault significance measure (FSM) index is designed to guide the selection of parameter in sparse random feature representation, which ensures the fault information richness of the required RCs. Finally, the feasibility and effectiveness of RSSCD are fully validated by simulation signals and two experimental cases. The results indicate that RSSCD has excellent decomposition performance and fault feature extraction ability.

Multi-view [formula omitted]-means clustering algorithm based on redundant and sparse feature learning

In recent years, multi-view clustering has been widely concerned and studied. Being its data generally has high dimension and noise, how to effectively deal with redundant features in data from various views and improve the clustering effect is an important issue in multi-view clustering. In this paper, a robust multi-view clustering algorithm Multi-view K-means clustering algorithm based on redundant and sparse feature learning (RSFMVKM) considering redundancy and sparse feature learning is proposed, which can effectively reduce the dimensions of data from different views. Firstly, the K-means algorithm is extended to multi-view clustering, and the importance of each views is represented by learning the weight of each view. Secondly, the sparse feature representation is obtained by applying the norm constraint to the projection matrix in a single view, and the cosine similarity is used to depict the redundancy matrix, which can remove the redundant information and improve the clustering effect. Moreover, the proposed algorithm can effectively avoid the generation of empty classes, and the clustering index matrix is discrete, so the clustering results can be obtained directly. Finally, the proposed algorithm is compared with six state-of-the-art multi-view clustering algorithms on six public datasets, and the experimental results show that the algorithm is effective.

RESTARTing the stabilized auditory image: How transient sampling of binaural information supports the emergence of stable auditory scenes

The perception of a spatial auditory scene involves the extraction and integration of multiple dynamic and unreliable sensory features (“cues”). Variation in each cue reflects the competing effects of multiple features of the auditory scene—e.g., relevant and irrelevant sound sources. Understanding which cue changes belong together—and to which objects—represents a fundamental challenge to the neural mechanisms of auditory scene perception. RESTART theory suggests a solution: transient, envelope-triggered sampling creates a temporally sparse representation of spatial features. Sparsity minimizes overlap between auditory objects and reduces ambiguity in the face of environmental distortions such as noise and reverberation. Envelope-triggered (“strobed”) sampling stabilizes the auditory image by emphasizing the most reliable cues. Here, we review the key evidence supporting RESTART theory, its neurocomputational mechanisms, and prior efforts to model them. [Work supported by US NIH R01-DC016643.]

Multi‐scale feature fusion pyramid attention network for single image dehazing

AbstractTexture and color distortion are common in existing learning‐based dehazing algorithms, and it is argued that one of the major reasons is that the shallow features of fog images are underutilized, and the deep features of fog images are insufficient for single image dehazing. In order to provide more texture and color information for image restoration, more shallow features need to be added in the process of image decoding. Therefore, a multi‐scale feature fusion pyramid attention network (PAN) for single image dehazing is proposed. In PAN, combined with the attention mechanism, a shallow and deep feature fusion (SDF) strategy is designed. SDF considers multi‐scale as well as channel‐level fusion to provide feature information under different receptive fields while also highlighting important channels, such as texture and color information. DC is designed as a latent space mapping module to learn a mapping relationship between the latent space representation of the hazy image at low resolution and the corresponding latent space representation of the haze‐free image. Additionally, network deconvolution (ND) and deformed convolution network (DCN) are introduced into PAN. The ND module can remove pixel‐wise and channel‐wise correlation of features, reduce data redundancy to obtain sparse representation of features, and speed up network convergence. The DCN module can use its adaptive receptive field to focus on the area of interest for calculation and play a role in texture feature enhancement. Finally, the perceptual loss is chosen as the regularization item of the loss function, which makes style features of the restored image closer to the real fog‐free image. Extensive experiments reveal that the proposed PAN outperforms other existing dehazing methods on real‐world and synthetic datasets.

Diffusion-weighted imaging-based radiomics for predicting 1-year ischemic stroke recurrence.

To investigate radiomics based on DWI (diffusion-weighted imaging) for predicting 1-year ischemic stroke recurrence. A total of 1,580 ischemic stroke patients were enrolled in this retrospective study conducted from January 2018 to April 2021. Demographic and clinical characteristics were compared between recurrence and non-recurrence groups. On DWI, lesions were segmented using a 2D U-Net automatic segmentation network. Further, radiomics feature extraction was done using the segmented mask matrix on DWI and the corresponding ADC map. Additionally, radiomics features were extracted. The study participants were divided into a training cohort (n = 157, 57 recurrence patients, and 100 non-recurrence patients) and a test cohort (n = 846, 28 recurrence patients, 818 non-recurrence patients). A sparse representation feature selection model was performed to select features. Further classification was accomplished using a recurrent neural network (RNN). The area under the receiver operating characteristic curve values was obtained for model performance. A total of 1,003 ischemic stroke patients (682 men and 321 women; mean age: 65.90 ± 12.44 years) were included in the final analysis. About 85 patients (8.5%) recurred in 1 year, and patients in the recurrence group were older than the non-recurrence group (P = 0.003). The stroke subtype was significantly different between recurrence and non-recurrence groups, and cardioembolic stroke (11.3%) and large artery atherosclerosis patients (10.3%) showed a higher recurrence percentage (P = 0.005). Secondary prevention after discharge (statins, antiplatelets, and anticoagulants) was found significantly different between the two groups (P = 0.004). The area under the curve (AUC) of clinical-based model and radiomics-based model were 0.675 (95% CI: 0.643-0.707) and 0.779 (95% CI: 0.750-0.807), respectively. With an AUC of 0.847 (95% CI: 0.821-0.870), the model that combined clinical and radiomic characteristics performed better. DWI-based radiomics could help to predict 1-year ischemic stroke recurrence.

Fault detection and diagnosis of the air handling unit via combining the feature sparse representation based dynamic SFA and the LSTM network

In recent years, slow feature analysis (SFA) has been successfully employed to deal with the air handling unit (AHU) system’s time-varying dynamic properties. However, since the derived slow features are the linear combinations of all original variables, the conventional SFA based methods may suffer from poor model interpretability and result in inaccurate fault detection and diagnosis (FDD) performance. To enhance the dynamic AHU system FDD effectiveness, this paper presents a novel feature sparse representation based SFA algorithm through imposing the sparsity on the slow features, which is called the sparse three-way data based dynamic SFA (STBDSFA). We make two contributions to propose the STBDSFA approach. One contribution is to infuse the feature sparse representation technique into the constructed SFA based monitoring model by performing the sparse restriction on the loading vector, which can effectively eliminate the meaningless variables’ coupling and select the key variable responsible for the fault detection. Before building the STBDSFA algorithm, the other contribution is to construct a new three-way data based dynamic SFA (TBDSFA) monitoring model to handle the AHU system’s two-directional dynamic properties. In the established TBDSFA model, the auto-regressive moving average exogenous (ARMAE) model is first adopted to reveal the variables’ auto-correlation relationships. Then, the multiway data analysis is applied to figure out the batch-wise dynamic property in multiple batch runs. After that, the dynamic SFA model is further set up to sufficiently tackle the time-wise dynamic nature within a batch run. To effectively diagnose the pattern of detected fault, another innovative work is to develop a long short-term memory (LSTM) based fault identification approach to classify the sparse slow features of fault snapshot dataset, owning to the LSTM’s superiority of coping with the dynamic time-varying nature of the sparse slow features. By comparing with some traditional and closely related FDD methods, the case study on the ASHRAE Research Project RP-1312 experimental datasets are performed to verify the performance and effectiveness of the proposed FDD scheme for the AHU system. To be specific, in comparison with the sparse principal component analysis, sparse dynamic SFA, kernel locality preserving projection and sparse three-way data based SFA approaches, the developed STBDSFA based monitoring method not only reveals no fault detection delays to alert the eleven test faults, but also achieves the highest average fault detection rates, i.e., 99.31% for the T2 statistic and 99.77% for the SPE statistic. Contrasted with the support vector machine, temporal convolutional network, convolutional neural network and deep belief network based classification algorithms, the suggested STLSTM based fault identification approach obtains the highest average fault diagnosis rate, i.e., 95.51% for the eleven fault snapshot datasets.

CT and MRI Medical Image Fusion Using Noise-Removal and Contrast Enhancement Scheme with Convolutional Neural Network.

Medical image fusion (MIF) has received painstaking attention due to its diverse medical applications in response to accurately diagnosing clinical images. Numerous MIF methods have been proposed to date, but the fused image suffers from poor contrast, non-uniform illumination, noise presence, and improper fusion strategies, resulting in an inadequate sparse representation of significant features. This paper proposes the morphological preprocessing method to address the non-uniform illumination and noise by the bottom-hat–top-hat strategy. Then, grey-principal component analysis (grey-PCA) is used to transform RGB images into gray images that can preserve detailed features. After that, the local shift-invariant shearlet transform (LSIST) method decomposes the images into the low-pass (LP) and high-pass (HP) sub-bands, efficiently restoring all significant characteristics in various scales and directions. The HP sub-bands are fed to two branches of the Siamese convolutional neural network (CNN) by process of feature detection, initial segmentation, and consistency verification to effectively capture smooth edges, and textures. While the LP sub-bands are fused by employing local energy fusion using the averaging and selection mode to restore the energy information. The proposed method is validated by subjective and objective quality assessments. The subjective evaluation is conducted by a user case study in which twelve field specialists verified the superiority of the proposed method based on precise details, image contrast, noise in the fused image, and no loss of information. The supremacy of the proposed method is further justified by obtaining 0.6836 to 0.8794, 0.5234 to 0.6710, and 3.8501 to 8.7937 gain for , CRR, and AG and noise reduction from 0.3397 to 0.1209 over other methods for objective parameters.

Improved Sparse Representation of Rolling Bearing Fault Features Based on Nested Dictionary

Improved Sparse Representation of Rolling Bearing Fault Features Based on Nested Dictionary

Sparse representation optimization of Gaussian mixed feature of image based on convolution neural network

With the rapid development of intelligent algorithm and image processing technology, the limitations of traditional image processing methods are more and more obvious. Based on this, this paper studies a new pattern of sparse representation optimization of image Gaussian mixture feature based on convolution neural network, and designs a sparse representation system model of vehicle detection image based on convolution neural network. The vehicle image data are collected from many aspects, and the convolution neural network is used for comprehensive analysis and evaluation. The model can extract the feature information of the vehicle detection image better by making the scheme of the real-time vehicle detection image and according to the image features and convolution neural network algorithm. The results show that the Gaussian mixture feature sparse representation optimization model based on convolution neural network has the advantages of high feasibility, high data accuracy and high response speed, which can enhance the processing efficiency of vehicle detection image and improve the utilization of local environmental information in the image.

Intelligent Diagnosis of Rotating Machinery Based on Optimized Adaptive Learning Dictionary and 1DCNN

In the intelligent fault diagnosis of rotating machinery, it is difficult to extract early weak fault impact features of rotating machinery under the interference of strong background noise, which makes the accuracy of fault identification low. In order to effectively identify the early faults of rotating machinery, an intelligent fault diagnosis method of rotating machinery based on an optimized adaptive learning dictionary and one-dimensional convolution neural network (1DCNN) is proposed in this paper. First of all, based on the original signal, a redundant dictionary with impact components is constructed by K-singular value decomposition (K-SVD), and the sparse coefficients are solved by an optimized orthogonal matching pursuit (OMP) algorithm. The sparse representation of fault impact features is realized, and the reconstructed signal with a concise fault impact feature structure is obtained. Secondly, the reconstructed signal is normalized, and the experimental dataset is divided into samples. Finally, the training set is input into the 1DCNN model for model training, and the test set is input into the trained model for classification and detection to complete the intelligent fault classification diagnosis of rotating machinery. This method is applied to the fault diagnosis of bearing data of Case Western Reserve University and worm gear reducer data of Shanghai University of Technology. Compared with other methods and models, the results show that the diagnosis method proposed in this paper can achieve higher diagnosis accuracy and better generalization ability than other diagnosis models under different datasets.

C-CNN: Contourlet Convolutional Neural Networks.

Extracting effective features is always a challenging problem for texture classification because of the uncertainty of scales and the clutter of textural patterns. For texture classification, spectral analysis is traditionally employed in the frequency domain. Recent studies have shown the potential of convolutional neural networks (CNNs) when dealing with the texture classification task in the spatial domain. In this article, we try combining both approaches in different domains for more abundant information and proposed a novel network architecture named contourlet CNN (C-CNN). The network aims to learn sparse and effective feature representations for images. First, the contourlet transform is applied to get the spectral features from an image. Second, the spatial-spectral feature fusion strategy is designed to incorporate the spectral features into CNN architecture. Third, the statistical features are integrated into the network by the statistical feature fusion. Finally, the results are obtained by classifying the fusion features. We also investigated the behavior of the parameters in contourlet decomposition. Experiments on the widely used three texture data sets (kth-tips2-b, DTD, and CUReT) and five remote sensing data sets (UCM, WHU-RS, AID, RSSCN7, and NWPU-RESISC45) demonstrate that the proposed approach outperforms several well-known classification methods in terms of classification accuracy with fewer trainable parameters.

Sparse Common Feature Representation for Undersampled Face Recognition

This work investigates the problem of undersampled face recognition (i.e., insufficient training data) encountered in practical Internet-of-Things (IoT) applications. Insufficient and uncertain samples captured by IoT devices may include background and facial disguise that makes face recognition more challenging than that with sufficient and reliable images. Many models work well in face recognition on a big data set, but when training data are insufficient, they achieve unsatisfactory performance. This work proposes a novel method named sparse common feature-based representation (SCFR) that provides a unique and stable result and completely avoids very time-consuming training required by a deep learning model. Specially, it constructs a common feature dictionary using both training and test images. Thereinto, a common feature is based on a discriminative common vector and learned by a Gaussian mixture model for both training and test images in a semisupervised learninig manner, which would reduce the difference among samples in each class. In the optimization, the latent indicator of test data is initialized by the estimated label. This can avoid learning invalid information and lead to good prototype images. A new variation dictionary characterizes variables that can be shared by different classes. Finally, this work adopts minimum reconstruction residuals to recognize test images, thus bringing about a substantial improvement in SCFR's performance. Extensive results on benchmark face databases demonstrate that the proposed method is better than the state-of-the-art methods handling undersampled face recognition.

Emotion recognition based on sparse representation of phase synchronization features

Emotion recognition based on Electroencephalogram (EEG) has attracted much attention in brain-computer interaction. However, most existing methods usually focus on amplitude and spectrum of the EEG signal, leading to sub-optimal performances due to the insufficiency in modelling the complex intrinsic information of neural integration. To address this issue, this paper proposes to capitalize on the largely neglected phase synchronization (PS) between EEG channels which reflects the intrinsic rhythmic interactions between different channels in neural integration. Specifically, this paper develops a simple and novel feature extraction method which calculates the PS based sparse representation features to analyze emotion states. First, the EEG phase synchronization indexes (PSI) of all channel pairs are estimated as features to distinguish different emotions, since certain topographical maps on PSI reveal specific emotion states. Then principal component analysis is performed to eliminate redundant and noisy features in PSI. Finally, Sparse Representation based Classification (SRC) furtherly emphasize emotion-related features and restrain useless features. For the benchmark affective EEG dataset DEAP, the proposed method based on no-overlapping EEG features achieve an average accuracy of 94.5%, 87.61%, and 67.04% for the classification tasks respectively on two, three and four emotions, demonstrating the superiority over state-of-the-art emotion classification methods.

Smart access control system by using sparse representation features

The access control system is used in massive scenarios to protect personal property safety, such as dormitory in campus, company office, and smart apartment. The recent access control system use either RFID card or biometric features, such as fingerprint, face. The RFID card may be lost and misused by other persons. The fingerprint features must be collected by touching a sensor, which is improper during the epidemic of Covid‐19. The face identification in access control system is effected by the mask, especially under the current epidemic situation. The research about partial occlusion face recognition is helpful to improve the existing face authentication for access control system. This paper establishes a smart access control system by using partial occlusion face recognition technology. First, the face is captured by a camera. Second, the collected faces are used to establish a face library for constructing low rank sparse model by using nuclear norm to measure the error matrix. Third, the sparse features of faces are used to train a neural network. The new personal face are verified by the learnt neural network model. The experimental results on a public partial occlusion face dataset demonstrate the effectiveness of the proposed access control system.

Multi-level cross-view consistent feature learning for person re-identification

Person re-identification plays an important role in searching for a specific person in a camera network with non-overlapping cameras. The most critical problem for re-identification is feature representation. In this paper, a multi-level cross-view consistent feature learning framework is proposed for person re-identification. First, local deep, LOMO and SIFT features are extracted to form multi-level features. Specifically, local features from the lower and higher layers of a convolutional neural network (CNN) are extracted, these features complement each other as they extract apparent and semantic properties. Second, an ID-based cross-view multi-level dictionary learning (IDB-CMDL) is carried out to obtain sparse and discriminant feature representation. Third, a cross-view consistent word learning is performed to get the cross-view consistent BoVW histograms from sparse feature representation. Finally, a multi-level metric learning fuses multiple BoVW histograms, and learns the sample distance in the subspace for ranking. Experiments on the public CUHK03, Market1501, and DukeMTMC-ReID datasets show results that are superior to many state-of-the-art methods for person re-identification.

Semisupervised Classification of Hyperspectral Image Based on Graph Convolutional Broad Network

Hyperspectral image (HSI) classification has attracted much attention in the field of remote sensing. However, the lack of sufficient labeled training samples is a huge challenge for HSI classification. To face this challenge, we propose a semisupervised HSI classification method based on graph convolutional broad network (GCBN). First, to avoid the underfitting problem caused by the insufficient linear sparse feature representation ability of broad learning system (BLS), graph convolution operation is applied to extract nonlinear and discriminative spectral-spatial features from the original HSI to replace the linear mapping features in the traditional BLS. Second, to solve the problem of insufficient model classification ability caused by limited labeled samples, the combinatorial average method (CAM) is proposed to use valuable paired samples to generate sample expansion set for GCBN model training. Third, BLS is used to perform broad expansion on spectral-spatial features extracted by GCN and extended by CAM, which further enhances the feature representation ability. Finally, the output weights can be easily calculated by the ridge regression theory. Experimental results on three real HSI datasets demonstrate the effectiveness of our proposed GCBN.