Sparse Feature Research Articles

A Learning-Free Method for Locomotion Mode Prediction by Terrain Reconstruction and Visual-Inertial Odometry.

This research introduces a novel, highly precise, and learning-free approach to locomotion mode prediction, a technique with potential for broad applications in the field of lower-limb wearable robotics. This study represents the pioneering effort to amalgamate 3D reconstruction and Visual-Inertial Odometry (VIO) into a locomotion mode prediction method, which yields robust prediction performance across diverse subjects and terrains, and resilience against various factors including camera view, walking direction, step size, and disturbances from moving obstacles without the need of parameter adjustments. The proposed Depth-enhanced Visual-Inertial Odometry (D-VIO) has been meticulously designed to operate within computational constraints of wearable configurations while demonstrating resilience against unpredictable human movements and sparse features. Evidence of its effectiveness, both in terms of accuracy and operational time consumption, is substantiated through tests conducted using open-source dataset and closed-loop evaluations. Comprehensive experiments were undertaken to validate its prediction accuracy across various test conditions such as subjects, scenarios, sensor mounting positions, camera views, step sizes, walking directions, and disturbances from moving obstacles. A comprehensive prediction accuracy rate of 99.00% confirms the efficacy, generality, and robustness of the proposed method.

Binocular Structured Light 3-D Reconstruction System for Low-Light Underwater Environments: Design, Modeling, and Laser-Based Calibration

The three dimensional (3D) perception of the low-light underwater environment has always been a major challenge which greatly limits the underwater operations. In this paper, an underwater active vision measurement system based on binocular structured light is designed to achieve high precision 3D reconstruction. Firstly, the fusion technology of binocular camera and laser addresses underwater optical attenuation and feature sparsity. Then, in order to avoid the huge inertia caused by the overall movement of the system, a laser scanner based on the mirror-galvanometer is used to accomplish static scanning of the scene. Subsequently, considering the influence of multiple media, underwater refraction models including monocular imaging model, binocular ranging model, and binocular polar curve constraint model are systematically proposed. What’s more, the conventional checkerboard-based passive visual calibration method is ineffective for low-light waters. Therefore, a simple calibration block is designed, and a new multi-objective laser-based calibration algorithm based on laser geometric constraints is proposed. Finally, the effectiveness of the system is verified by analyzing the 3D reconstruction results of underwater objects.

Which Target to Focus on: Class-Perception for Semantic Segmentation of Remote Sensing

Deep Learning-based (DL) methods have dominated the task of semantic segmentation of remote sensing images. However, the sizes of different objects vary widely, and there is a great deal of label-noise due to the inevitable shadows. Therefore, there is an urgent need for a method that can precisely handle complex ground data. In this paper, we propose an Inter-Class Enhanced Network (ICEN) for representing features of varying sizes. It comprises two branches: Sparse Representation Network (SPN) and Feature Extraction Network (FEN). Then, a Class-Perception Block is inserted between the two branches to instruct the SPN’s low-level semantic features to be merged into the deeper network. Such a block can reduce label-noise in remote sensing image segmentation. In addition, the proposed EIRI provides a more precise classification process for target edges containing many misclassified points without requiring excessive computational overhead. The experimental results of our proposed Class-Perception Network (C-PNet) achieve competitive performance on the Vaihingen, Potsdam, LoveDA, and UAVid datasets.

Sparse control-inspired features generation for Schizophrenia diagnosis⋆

In this paper, a recent control-oriented features generation method is used to diagnose schizophrenia using electroencephalogram (EEG) signals. The methodology has already been used for the diagnosis of Parkinson's disease with competitive results. The method is directly inspired by the functioning of the brain and is mainly based on optimal control theory and sparse optimisation. An appealing feature in the proposed solution is that it allows to combine both frequency and temporal-related aspects of the signal which are known to be detrimental in this context. The proposed solution is evaluated on a publicly available dataset that includes 81 subjects, of which 49 suffer from schizophrenia and 32 are healthy. Results show that by mean of only one extracted feature, fed to a linear discriminant analysis (LDA) classifier, high accuracy separation is obtained. Several validity tests were carried out to assess the statistical relevance of the findings.

Deep Feature Enhancement Method for Land Cover With Irregular and Sparse Spatial Distribution Features: A Case Study on Open-Pit Mining

Land cover classification in mining areas (LCMA) is essential for the environmental assessment of mines and plays a crucial role in their sustainable development. The shapes of mine land occupation elements are irregular, and the overall proportion of their area is relatively small. Therefore, their features may be easily lost during feature extraction, which limits the interpretation accuracy in mining areas. This study attempts to address these issues. We propose a model named EG-UNet to enhance the features of elements with few samples and to capture long-range information. The proposed EG-UNet includes two main modules: 1) The edge feature enhancement module: the edges of elements of mine land occupation contain more information than other spatial locations. Hence, during the feature extraction of elements, a Sobel operator is used to extract the object boundary, which increases the weight of these features before the pooling operation for their preservation. 2) The long-range information extraction module: long-range information helps extract tiny objects, such as dumping grounds in the mining area. We present a graph convolution network (GCN) to capture the long-range features and apply convolutional neural networks to learn the graph construction. A total of 10 deep-learning networks were compared using the LCMA semantic segmentation dataset. Our model exhibited the best performance, especially in classifying classes with few samples. Further, to evaluate the general ability of EG-UNet, a benchmark-Gaofen Image Dataset (GID) was used, and the result still reflected the superiority of our method.

Dissecting Tensor Cores via Microbenchmarks: Latency, Throughput and Numeric Behaviors

Tensor Cores have been an important unit to accelerate Fused Matrix Multiplication Accumulation (MMA) in all NVIDIA GPUs since Volta Architecture. To program Tensor Cores, users have to use either legacy wmma APIs or current mma APIs. Legacy wmma APIs are more easy-to-use but can only exploit limited features and power of Tensor Cores. Specifically, wmma APIs support fewer operand shapes and can not leverage the new sparse matrix multiplication feature of the newest Ampere Tensor Cores. However, the performance of current programming interface has not been well explored. Furthermore, the computation numeric behaviors of low-precision floating points (TF32, BF16, and FP16) supported by the newest Ampere Tensor Cores are also mysterious. In this paper, we explore the throughput and latency of current programming APIs. We also intuitively study the numeric behaviors of Tensor Cores MMA and profile the intermediate operations including multiplication, addition of inner product, and accumulation. All codes used in this work can be found in https://github.com/sunlex0717/DissectingTensorCores.

A Semantics-Geometry Framework for Road Extraction From Remote Sensing Images

Road extraction from remote sensing images in very high resolution is important for autonomous driving and road planning. Compared with large-scale objects, roads are smaller, winding, and likely to be covered by buildings’ shadows, causing deep convolutional neural networks (DCNNs) to be difficult to identify roads. The paper proposes a semantics-geometry framework (SGNet) with a two-branch backbone, i.e., semantics-dominant branch and geometry-dominant branch. The semantics-dominant branch inputs images to predict dense semantic features, and the geometry-dominant branch takes images to generate sparse boundary features. Then, dense semantic features and boundary details generated by two branches are adaptively fused. Further, by utilizing affinity between neighborhood pixels, a feature refinement module is proposed to refine textures and road details. We evaluate the SGNet on the Ottawa road dataset. Experiments show that the SGNet outperforms other competitors on the road extraction task. Codes is available at https://github.com/qiuluyi/SGNet.

Optical and SAR Image Dense Registration Using a Robust Deep Optical Flow Framework

The coregistration of optical and synthetic aperture radar (SAR) imageries is the bottleneck in exploring the complementary information from the two multimodal datasets. The difficulties lie in not only the complex radiometric relationship between them, but also the distinct geometrical models of the optical and SAR imaging systems, which cause it nontrivial to explicitly depict the spatial relationship between the corresponding image regions when elevation fluctuations exist. This article aims to investigate the optical flow technique for the pixelwise dense registration of the high-resolution optical and SAR images, so as to get rid of the outlier removal and geometric mapping procedures, which have to be conducted in the classical image registration approaches that are based on sparse feature point matching. Herein, a deep optical flow framework is designed. First, a dilated feature concatenation method is proposed to enhance the discriminability of the pixelwise features for similarity measurement. An effective network training strategy is used, based on a smoothed flow loss, and also a training dataset that contains simulated elevation fluctuations. Second, we propose a self-supervised optical flow fine-tuning strategy. It incorporates the strength of the blockwise matching approach, which produces better matching precision, into the proposed pixelwise matching procedure. In this way, the accuracy of the optical-SAR dense registration is substantially improved. Extensive experiments conducted on the 1-m resolution optical-SAR image pairs of different land-cover types and distinct topographic conditions indicate that the proposed optical-SAR optical flow network -Ft framework is quite robust, and has the potential to perform the optical-SAR image dense registration in practical applications. The Python code of the proposed deep optical flow network will be made available.

Saliency-Based SAR Target Detection via Convolutional Sparse Feature Enhancement and Bayesian Inference

Traditional synthetic aperture radar (SAR) target detection methods use matched filtered SAR images as input, and the detection performance is restricted due to the high sidelobes and speckle noise of these images. Sparse SAR imaging methods developed in recent years provide the advantages of reducing sidelobes, noise, and clutter. The imaging results obtained with these methods could help improve the SAR target detection performance. In this article, to improve the target detection performance using sparse SAR images as input, we proposed a convolutional sparse feature enhancement method to meet the needs of Bayesian saliency detection. The proposed Bayesian saliency joint target detection method comprised the following three steps: first, to obtain sparse SAR images with continuous contours and fewer holes in the target area, we proposed a convolutional L1 sparse regularization method. Second, a regularization parameter optimization method was derived to quickly obtain optimal regularization parameters for saliency detection. Finally, target detection results were obtained through a superpixel-based Bayesian saliency joint detector. Extensive experiments verified that the proposed method could improve the SAR target detection accuracy in complex backgrounds.

Sequential Anomaly Detection Against Demodulation Reference Signal Spoofing in 5G NR

In fifth generation (5G) new radio (NR), the demodulation reference signal (DMRS) is employed for channel estimation as part of coherent demodulation of the physical uplink shared channel. However, DMRS spoofing poses a serious threat to 5G NR since inaccurate channel estimation will severely degrade the decoding performance. In this correspondence, we propose to exploit the spatial sparsity structure of the channel to detect the DMRS spoofing, which is motivated by the fact that the spatial sparsity structure of the channel will be significantly impacted if the DMRS spoofing happens. We first extract the spatial sparsity structure of the channel by solving a sparse feature retrieval problem, then propose a sequential sparsity structure anomaly detection method to detect DMRS spoofing. In simulation experiments, we exploit clustered delay line based channel model from 3GPP standards for verifications. Numerical results show that our method outperforms both the subspace dimension based and energy detector based methods.

An Efficient Visual SfM Framework Using Planar Markers

Structure-from-Motion (SfM) is a prevalent solution for vision measurement. Traditional SfM methods rely on sparse feature matching to establish visual associations between different views. However, both the incremental-based and global-based SfM show sensitiveness to outliers generated by existing hand-crafted features under challenging environments. Planar marker-assisted SfM is an alternative solution that employs artificial landmarks to help robust pose estimation. Nevertheless, using planar markers also exists the planar pose ambiguity problem, which means there may be more than one reasonable solution for pose estimation; it is a great challenge for camera pose estimation. This paper proposes an efficient planar marker-assisted global SfM method and resolves the planar pose ambiguity problem by examining the rotation consistency across multiple views. In particular, we construct two complementary graphs and present a graph filter method to filter out the views with pose ambiguity according to our designed rotation consistency checking. Afterward, we perform sparse feature matching following the proposed graph structure and make a pose-fixed bundle adjustment for high-quality 3D reconstruction. Extensive experiments have been conducted on synthetic and real datasets, and the results verify the effectiveness of the proposed method. Our method outperforms several state-of-the-art methods with encouraging efficiency, especially in challenging scenes.

A Lightweight Real-Time 3D LiDAR SLAM for Autonomous Vehicles in Large-Scale Urban Environment

For autonomous vehicles, real-time localization and mapping in the unknown environment is very important. In this paper, a fast and lightweight 3D LiDAR simultaneously localization and mapping (SLAM) is presented for the localization of autonomous vehicles in large-scale urban environments. A novel encoding approach based on depth information is proposed to encode unordered point clouds with various resolutions, which avoids missing dimensional information in the projection of point clouds onto a 2D plane. Principal components analysis (PCA) is modified by dynamically selecting neighborhood points according to the encoded depth information to fit the local plane with less time consuming. The threshold and the number of feature points are adaptive according to distance intervals, results in sparse feature points extracted and uniformly distributed in the three-dimensional space. The extracted significant feature points improve the accuracy of the odometer and speed up the alignment of the point cloud. The effectiveness and robustness of the proposed algorithm are verified on the KITTI odometry benchmark and MVSECD. A fast runtime of 21 ms is obtained for the odometer estimation. Compared to several typical state-of-the-art methods on the KITTI odometry benchmark, the proposed approach reduces translation errors by at least 19% and rotation errors by 7.1%.

A Novel Neighborhood Rough Set-Based Feature Selection Method and Its Application to Biomarker Identification of Schizophrenia.

Feature selection can disclose biomarkers of mental disorders that have unclear biological mechanisms. Although neighborhood rough set (NRS) has been applied to discover important sparse features, it has hardly ever been utilized in neuroimaging-based biomarker identification, probably due to the inadequate feature evaluation metric and incomplete information provided under a single-granularity. Here, we propose a new NRS-based feature selection method and successfully identify brain functional connectivity biomarkers of schizophrenia (SZ) using functional magnetic resonance imaging (fMRI) data. Specifically, we develop a new weighted metric based on NRS combined with information entropy to evaluate the capacity of features in distinguishing different groups. Inspired by multi-granularity information maximization theory, we further take advantage of the complementary information from different neighborhood sizes via a multi-granularity fusion to obtain the most discriminative and stable features. For validation, we compare our method with six popular feature selection methods using three public omics datasets as well as resting-state fMRI data of 393 SZ patients and 429 healthy controls. Results show that our method obtained higher classification accuracies on both omics data (100.0%, 88.6%, and 72.2% for three omics datasets, respectively) and fMRI data (93.9% for main dataset, and 76.3% and 83.8% for two independent datasets, respectively). Moreover, our findings reveal biologically meaningful substrates of SZ, notably involving the connectivity between the thalamus and superior temporal gyrus as well as between the postcentral gyrus and calcarine gyrus. Taken together, we propose a new NRS-based feature selection method that shows the potential of exploring effective and sparse neuroimaging-based biomarkers of mental disorders.

GeSANet: Geospatial-Awareness Network for VHR Remote Sensing Image Change Detection

The characteristics of very high resolution (VHR) remote sensing images (RSIs) have higher spatial resolution inherently, and are easier to obtain globally compared with hyperspectral images (HSIs), making it possible to detect small-scale land cover changes in multiple applications. RSI change detection (RSI-CD) based on deep learning has been paid attention to and become a frontier research field in recent years, and is currently facing two challenging problems: The first is high dependence on registration between bi-temporal images caused by high spatial resolution; The other is high pseudo-change information response caused by low spectral resolution. In order to address the above-mentioned two problems, a novel RSI-CD framework called Geospatial-Awareness Network (GeSANet) based on the geospatial Position Matching Mechanism (PMM) with multi-level adjustment and the geo-spatial Content Reasoning Mechanism (CRM) with diverse pseudo-change information filtering is proposed. First of all, the PMM assigns independent two-dimensional offset coordinates to each position in the previous temporal image, afterwards, bilinear interpolation is employed to obtain the subpixel feature value after the offset, and the sparse results based on the difference are transmitted to the next level prediction to realize multi-level geospatial correction. The CRM extracts global features from the corrected sparse feature map in terms of dimensions, implementing effective discriminant feature extraction on basis of the original feature map in a stepwise refinement manner through the cross-dimension exchange mechanism, to filter out various pseudo-change information as well as maintain real change information. Comparison experiments with five recent SOTA methods are carried out on two popular datasets with diverse changes, the results show that the proposed method has good robustness and validity for multi-temporal RSI-CD. In particular, it has a strong comparative advantage in detecting small entity changes and edge details. The source code of the proposed framework can be downloaded from https://github.com/zxylnnu/GeSANet.

Sparse Feature Points Extraction-Based Localization With Partial Information Loss in UWSNs

This paper investigates the problem of localization of mobile nodes in the marine environment under conditions of information deficiency. To solve this problem, sparse feature points (SFPs) extraction-based time of flight (ToF) minimum residual localization algorithm and localization based on the prior information of the target under partial information loss are proposed. First, the SFPs extraction method is adopted to fit the sound speed profile (SSP). Based on this, the SFPs extraction-based ToF computational model is proposed. The coordinates of the nodes to be located are calculated by ToF measurement and the ToF model for minimum residual localization. Second, in the case of information loss, particle ranges are generated using the target history a prior information. This is combined with the received node localization information to locate the node by the proposed target prior information-based localization method. Finally, the results of the simulation experiments show that the proposed method achieves a more detailed description of the SSP characteristics. The localization error of the proposed method is reduced fivefold compared with other methods under the condition of information loss, which is more in line with the spatial characteristics of the underwater environment.

Blind Goal-Oriented Massive Access for Future Wireless Networks

Emerging communication networks are envisioned to support massive wireless connectivity of heterogeneous devices with sporadic traffic and diverse requirements in terms of latency, reliability, and bandwidth. Providing multiple access to an increasing number of uncoordinated users and sharing the limited resources become essential in this context. In this work, we revisit the random access (RA) problem and exploit the continuous angular group sparsity feature of wireless channels to propose a novel RA strategy that provides low latency, high reliability, and massive access with limited bandwidth resources in an all-in-one package. To this end, we first design a reconstruction-free goal-oriented optimization problem, which only preserves the angular information required to identify the active devices. To solve this, we propose an alternating direction method of multipliers (ADMM) and derive closed-form expressions for each ADMM step. Then, we design a clustering algorithm that assigns the users in specific groups from which we can identify active stationary devices by their angles. For mobile devices, we propose an alternating minimization algorithm to recover their data and their channel gains simultaneously, which allows us to identify active mobile users. Simulation results show significant performance gains in terms of active user detection and false alarm probabilities as compared to state-of-the-art RA schemes, even with limited number of preambles. Moreover, unlike prior work, the performance of the proposed blind goal-oriented massive access does not depend on the number of devices.

Application of Compressive Sensing (CS) to Wide-Band Cognitive Radio signals

Compressive Sensing (CS) is a digital signal processing developed theory that encloses the signal sampling and compression, based on the sparsity characteristics of signal. This can decrease sampling rate, so reduce computational complexity of the system without degrading the performance of the system. This paper describes the theoretical frame and a few key technical, then illustrates the application of compressed sensing theory to wide-band cognitive radio signals. Spectrum sensing is a critical issue in wide-band Cognitive Radio (CR) networks as it faces hard challenges such as high hardware cost, complexity, sampling rate and processing speed. Thus, this paper shows that Compressive Sensing could be exploited in wide-band Cognitive Radio networks to solve the spectrum sensing problems mentioned above.

Generalized Soft-Root-Sign Based Robust Sparsity-Aware Adaptive Filters

Robust adaptive filters utilizing hyperbolic cosine and correntropy functions have been successfully employed in non-Gaussian noisy environments. However, these filters suffer from high steady-state misalignment due to significant weight update in the presences of outliers. In addition, several practical systems exhibit sparse characteristics, which is not taken into account by these filters. In this paper, a generalized soft-root-sign (GSRS) function is proposed and the corresponding GSRS adaptive filter is designed. The proposed GSRS provides negligible weight update in the occurrence of large outliers and thereby results in lower steady-state misalignment. To further improve modelling performance for sparse systems and to achieve robustness, sparsity-aware GSRS algorithms are also developed in this paper. The bound on learning rate and the computational complexity of proposed algorithm is also investigated. Simulation studies confirmed the improved convergence characteristics achieved by the proposed algorithms over existing algorithms.

Block Sparse Bayesian Learning Based Joint User Activity Detection and Channel Estimation in Grant-Free MIMO-NOMA

In the massive machine type of communication (mMTC), grant-free non-orthogonal multiple access (NOMA) is receiving more and more attention because it can skip the complex grant process to allocate non-orthogonal resources to serve more users. To address the limited wireless resources and substantial connection challenges, combining grant-free NOMA and multiple-input multiple-output (MIMO) is crucial to further improve the system’s capacity. In the grant-free MIMO-NOMA system, the base station should obtain the relevant information of the user before data detection. Thus, user activity detection (UAD) and channel estimation (CE) are two problems that should be solved urgently. In this paper, we fully consider the sparse characteristics of signals and the spatial correlation between multiple antennas in the grant-free MIMO-NOMA system. Then, we propose a spatial correlation block sparse Bayesian learning (SC-BSBL) algorithm to address the joint UAD and CE problems. First, by fully mining the block sparsity of signals in the grant-free MIMO-NOMA system, we model the joint UAD and CE problem as a three-dimensional block sparse signal recovery problem. Second, we derive the cost function based on the hierarchical Bayesian theory and spatial correlation. Finally, to estimate the channel and the set of active users, we optimize the cost function with fast marginal likelihood maximization. The simulation results indicate that, compared with the existing algorithms, SC-BSBL can always fully use the signal sparsity and spatial correlation to accurately complete UAD and CE under various user activation probabilities, SNRs, and the number of antennas. The normalized mean square error of CE can be reduced to 0.01, and the UAD error rate can be less than 10−5.

Tree Reconstruction Using Topology Optimisation

Generating accurate digital tree models from scanned environments is invaluable for forestry, agriculture, and other outdoor industries in tasks such as identifying fall hazards, estimating trees’ biomass and calculating traversability. Existing methods for tree reconstruction rely on sparse feature identification to segment a forest into individual trees and generate a branch structure graph, limiting their application to easily separable trees and uniform forests. However, the natural world is a messy place in which trees present with significant heterogeneity and are frequently encroached upon by the surrounding environment. We present a general method for extracting the branch structure of trees from point cloud data, which estimates the structure of trees by adapting the methods of structural topology optimisation to find the optimal material distribution to interpolate the input data. We present the results of this optimisation over a wide variety of scans, and discuss the benefits and drawbacks of this novel approach to tree structure reconstruction. Our method generates detailed and accurate tree structures, with a mean Surface Error (SE) of 15 cm over 13 diverse tree datasets.