Reduce Information Loss Research Articles

TranSkeleton: Hierarchical Spatial–Temporal Transformer for Skeleton-Based Action Recognition

In skeleton-based action recognition, it has been a dominant paradigm to extract motion features with temporal convolution and model spatial correlations with graph convolution. However, it’s difficult for temporal convolution to capture long-range dependencies effectively. Meanwhile, commonly used multi-branch graph convolution leads to high complexity. In this paper, we propose TranSkeleton, a powerful Transformer framework which neatly unifies the spatial and temporal modeling of skeleton sequences. For temporal modeling, we propose a novel partition-aggregation temporal Transformer. It works with hierarchical temporal partition and aggregation, and can capture both long-range dependencies and subtle temporal structures effectively. A difference-aware aggregation approach is designed to reduce information loss during temporal aggregation. For spatial modeling, we propose a topology-aware spatial Transformer which utilizes the prior information of human body topology to facilitate spatial correlation modeling. Extensive experiments on two challenging benchmark datasets demonstrate that TranSkeleton notably outperforms the state of the arts.

Hippocampus Segmentation Method Applying Coordinate Attention Mechanism and Dynamic Convolution Network

Precisely segmenting the hippocampus from the brain is crucial for diagnosing neurodegenerative illnesses such as Alzheimer’s disease, depression, etc. In this research, we propose an enhanced hippocampus segmentation algorithm based on 3D U-Net that can significantly increase hippocampus segmentation performance. First, a dynamic convolution block is designed to extract information more comprehensively in the steps of the 3D U-Net’s encoder and decoder. In addition, an improved coordinate attention algorithm is applied in the skip connections step of the 3D U-Net to increase the weight of the hippocampus and reduce the redundancy of other unimportant location information. The algorithm proposed in this work uses soft pooling methods instead of max pooling to reduce information loss during downsampling steps. The datasets employed in this research were obtained from the MICCAI 2013 SATA Challenge (MICCAI) and the Harmonized Protocol initiative of the Alzheimer’s Disease Neuroimaging Initiative (HarP). The experimental results on the two datasets prove that the algorithm proposed in this work outperforms other commonly used segmentation algorithms. On the HarP, the dice increase by 3.52%, the mIoU increases by 2.65%, and the F1 score increases by 3.38% in contrast to the baseline. On the MICCAI, the dice, the mIoU, and the F1 score increase by 1.13%, 0.85%, and 1.08%, respectively. Overall, the proposed model outperforms other common algorithms.

Doctor selection based on aspect-based sentiment analysis and neutrosophic TOPSIS method

Background:Physician-rating sites have become a convenient platform for patients to choose doctors. However, selecting an appropriate doctor through numerous online reviews is challengeable for patients. Although studies show that patients have different preferences for the aspects of healthcare services, existing doctor ranking methods rarely consider such preference information. Besides, they seldom handle the neutral sentiment information in patient reviews. Methods:To better assist patients in doctor selection, we propose a novel decision-making method that combines aspect-based sentiment analysis, single-valued neutrosophic sets and an extended Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) method. The method utilizes Word2Vec to construct a feature dictionary of patient satisfaction. A rule-based approach is employed to extract the aspects and the related sentiments from patients’ text reviews. Moreover, it takes advantage of single-valued neutrosophic sets to address the positive, neutral and negative sentiment information. On this basis, we rank the doctors with an extended TOPSIS method considering the patient’s attitudinal character and the preference information. Results:A case study on a review dataset demonstrates the stability and flexibility of the method. The comparative analysis reveals that our method can efficiently reduce information loss and it is more practical than existing methods. Conclusion:Considering patients’ different preferences for the aspects of healthcare services, the method can better provide healthcare decision support for patients.

A convolutional spiking neural network with adaptive coding for motor imagery classification

Motor imagery (MI) signal classification is crucial for brain-computer interfaces (BCI). The third-generation neural network, spiking neural network (SNN), has rich neurodynamic properties in the spatiotemporal domain, and therefore it is more suitable for processing EEG signals. However, the feature extraction capability of the SNN previously applied to MI signal classification is limited by its structure, and the model’s classification accuracy is not comparable to the state-of-the-art algorithms. In this paper, we propose a spiking neural network model called SCNet, which combines the feature extraction capability of CNN with the biological interpretability of SNN, making the model structurally closer to the biological neuronal dynamical system and improving the classification accuracy. SCNet reduces information loss by adaptive coding with learnability and solves the training difficulties of spiking neural networks by surrogate gradient learning. We evaluated the performance of the proposed SCNet on three typically representative motor imagery datasets. The validation shows that the model outperforms state-of-the-art SNN-based MI classification methods and various ANN and machine learning methods. The experimental results demonstrate the generality and effectiveness of the proposed motor imagery EEG signal classification model. Better classification results can be obtained by designing a well-structured spiking neural network.

A Small Object Detection Network Based on Multiple Feature Enhancement and Feature Fusion

Due to the small size, high resolution, and complex background, small object detection has become a difficult point in computer vision. Making full use of high-resolution features and reducing information loss in the process of information propagation is of great significance to improve small object detection. In this article, to achieve the above two points, this work proposes a small object detection network based on multiple feature enhancement and feature fusion based on RetinaNet (MFEFNet). First, this work designs a densely connected dilated convolutions to adequately extract high-resolution features from C2. Then, this work utilizes subpixel convolution to avoid the loss of channel information caused by channel dimension reduction in the lateral connection. Finally, this article introduces a bidirectional fusion feature pyramid structure to shorten the propagation path of high-resolution features and reduce the loss of high-resolution features. Experiments show that our proposed MFEFNet achieves stable performance gains in object detection task. Specifically, the improved method improves RetinaNet from 34.4AP to 36.2AP on the challenging MS COCO dataset, and especially achieves excellent results in small object detection with an improvement of 2.9%.

Graph-Clustering Anonymity Privacy Protection Algorithm With Fused Distance-Attributes

Clustering anonymity is a common social network data privacy protection scheme, which is based on graph-clustering. Many existing graph clustering methods mainly focus on the relationship between the structure and attributes of nodes, and the difference between them due to the metric usually causes the problem of poor clustering results. To address the shortcomings in the above graph-clustering methods, a graph-clustering anonymity method implemented with fused distance-attributes (GCA-DA) is proposed. Firstly, the algorithm quantifies the distance and attribute similarity between nodes separately and balances the metric differences between them to calculate the integrated similarity. Then all the nodes in the graph are clustered into clusters according to the integrated similarity between two nodes, each of which contains no fewer than k nodes. Finally, all the clusters are anonymized. In this method, the attribute generalization for every cluster can prevent attacks by the background knowledge of structure and attributes. In addition, the attributes are divided into numerical and non-numerical attributes to measure them separately, therefore can maintain the usability of the data better. Experiment results demonstrate the effectiveness in improving the quality of clustering and reducing information loss.

Multiple criteria decision analytic methods in management with T-spherical fuzzy information.

With a focus on T-spherical fuzzy (T-SF) sets, the aim of this paper is to create a split-new appraisal mechanism and an innovative decision analytic method for use with multiple-criteria assessment and selection in uncertain situations. The T-SF frame is the latest recent advancement in fuzzy settings and uses four facets (consisting of membership grades of positivity, neutrality, negativity, and refusal) to elucidate complex uncertainties, thereby evidently reducing information loss, in anticipation of fully manifesting indistinct and equivocal information. This paper adds to the body of knowledge regarding multiple criteria choice modeling by raising T-SF correlation-oriented measurements connected to the fixed and displaced ideal/anti-ideal benchmarks and by creating an approachable appraisal mechanism for advancing a T-SF decision analytic methodology. Consider, in particular, the performance ratings of available options in terms of judging criteria under the T-SF type of uncertainties. This research gives correlation-oriented measurements focusing on two varieties of maximum and square root functions in T-SF situations, which serve as a solid foundation for an efficacious appraisal mechanism from two views of anchored judgments corresponding to the fixed and displaced benchmarks. The T-SF Minkowski distance index is generated to integrate the outranking and outranked identifiers relying on correlation-oriented measurements for figuring out the local outranking and outranked indices. The T-SF decision analytic procedures are constructed using a new appraisal significance index that is founded on certain valuable insights of correlation-oriented maximizing and minimizing indices as well as global outranking and outranked indices. Additionally, a concrete location selection dilemma is dealt with in this research to showcase the applicability and efficiency of the suggested T-SF decision analytic methodology. Sensitivity analyses and comparative studies are carried out to investigate substantial modifications in pertinent parameters and to confirm the robustness of the predominance relationships among the available options. The suggested approaches are adaptable, flexible, and reliable, according to the application outcomes and comparison findings. This research provides four scientific contributions: (1) the utilization of T-SF correlation coefficients as the basis for prioritization analysis involving multiple criteria assessments, (2) the evolution of the T-SF Minkowski distance index to model outranking decision-making processes, (3) the creation of a reliable appraisal mechanism based on T-SF correlation-oriented measurements for intelligent decision support, and (4) the advancement of computational tools and procedures (e.g., correlation-oriented maximizing and minimizing indices, global outranking and outranked indices, and appraisal significance indices) to perform the decision analytic procedure in T-SF settings. In terms of managerial implications, the solution findings support the employment of the fixed ideal/anti-ideal benchmarking mechanism, as its measurements and indices are easy to operate and suitably sensitive. Next, in practical implementations of the T-SF decision analytic procedure, it is advised to utilize the T-SF Manhattan distance index for calculating convenience. Finally, the T-SF decision analytic techniques offer fundamental ideas and measurements appropriate for manipulating T-SF information in complex decision situations, thereby increasing the application potential in the area of decision-making with information uncertainty.

ACTNet: A Dual-Attention Adapter with a CNN-Transformer Network for the Semantic Segmentation of Remote Sensing Imagery

In recent years, the application of semantic segmentation methods based on the remote sensing of images has become increasingly prevalent across a diverse range of domains, including but not limited to forest detection, water body detection, urban rail transportation planning, and building extraction. With the incorporation of the Transformer model into computer vision, the efficacy and accuracy of these algorithms have been significantly enhanced. Nevertheless, the Transformer model’s high computational complexity and dependence on a pre-training weight of large datasets leads to a slow convergence during the training for remote sensing segmentation tasks. Motivated by the success of the adapter module in the field of natural language processing, this paper presents a novel adapter module (ResAttn) for improving the model training speed for remote sensing segmentation. The ResAttn adopts a dual-attention structure in order to capture the interdependencies between sets of features, thereby improving its global modeling capabilities, and introduces a Swin Transformer-like down-sampling method to reduce information loss and retain the original architecture while reducing the resolution. In addition, the existing Transformer model is limited in its ability to capture local high-frequency information, which can lead to an inadequate extraction of edge and texture features. To address these issues, this paper proposes a Local Feature Extractor (LFE) module, which is based on a convolutional neural network (CNN), and incorporates multi-scale feature extraction and residual structure to effectively overcome this limitation. Further, a mask-based segmentation method is employed and a residual-enhanced deformable attention block (Deformer Block) is incorporated to improve the small target segmentation accuracy. Finally, a sufficient number of experiments were performed on the ISPRS Potsdam datasets. The experimental results demonstrate the superior performance of the model described in this paper.

Fine-grained similarity semantic preserving deep hashing for cross-modal retrieval

Cross-modal hashing methods have received wide attention in cross-modal retrieval owing to their advantages in computational efficiency and storage cost. However, most existing deep cross-modal hashing methods cannot employ both intra-modal and inter-modal similarities to guide the learning of hash codes and ignore the quantization loss of hash codes, simultaneously. To solve the above problems, we propose a fine-grained similarity semantic preserving deep hashing (FSSPDH) for cross-modal retrieval. Firstly, this proposed method learns different hash codes for different modalities to preserve the intrinsic property of each modality. Secondly, the fine-grained similarity matrix is constructed by using labels and data features, which not only maintains the similarity between and within modalities. In addition, quantization loss is used to learn hash codes and thus effectively reduce information loss caused during the quantization procedure. A large number of experiments on three public datasets demonstrate the advantage of the proposed FSSPDH method.

MSFANet: Multiscale Fusion Attention Network for Road Segmentation of Multispectral Remote Sensing Data

With the development of deep learning and remote sensing technologies in recent years, many semantic segmentation methods based on convolutional neural networks (CNNs) have been applied to road extraction. However, previous deep learning-based road extraction methods primarily used RGB imagery as an input and did not take advantage of the spectral information contained in hyperspectral imagery. These methods can produce discontinuous outputs caused by objects with similar spectral signatures to roads. In addition, the images obtained from different Earth remote sensing sensors may have different spatial resolutions, enhancing the difficulty of the joint analysis. This work proposes the Multiscale Fusion Attention Network (MSFANet) to overcome these problems. Compared to traditional road extraction frameworks, the proposed MSFANet fuses information from different spectra at multiple scales. In MSFANet, multispectral remote sensing data is used as an additional input to the network, in addition to RGB remote sensing data, to obtain richer spectral information. The Cross-source Feature Fusion Module (CFFM) is used to calibrate and fuse spectral features at different scales, reducing the impact of noise and redundant features from different inputs. The Multiscale Semantic Aggregation Decoder (MSAD) fuses multiscale features and global context information from the upsampling process layer by layer, reducing information loss during the multiscale feature fusion. The proposed MSFANet network was applied to the SpaceNet dataset and self-annotated images from Chongzhou, a representative city in China. Our MSFANet performs better over the baseline HRNet by a large margin of +6.38 IoU and +5.11 F1-score on the SpaceNet dataset, +3.61 IoU and +2.32 F1-score on the self-annotated dataset (Chongzhou dataset). Moreover, the effectiveness of MSFANet was also proven by comparative experiments with other studies.

LASSO and attention-TCN: a concurrent method for indoor particulate matter prediction.

Long time exposure to indoor air pollution environments can increase the risk of cardiovascular and respiratory system damage. Most previous studies focus on outdoor air quality, while few studies on indoor air quality. Current neural network-based methods for indoor air quality prediction ignore the optimization of input variables, process input features serially, and still suffer from loss of information during model training, which may lead to the problems of memory-intensive, time-consuming and low-precision. We present a novel concurrent indoor PM prediction model based on the fusion model of Least Absolute Shrinkage and Selection Operator (LASSO) and an Attention Temporal Convolutional Network (ATCN), together called LATCN. First, a LASSO regression algorithm is used to select features from PM1, PM2.5, PM10 and PM (>10) datasets and environmental factors to optimize the inputs for indoor PM prediction model. Then an Attention Mechanism (AM) is applied to reduce the redundant temporal information to extract key features in inputs. Finally, a TCN is used to forecast indoor particulate concentration in parallel with inputting the extracted features, and it reduces information loss by residual connections. The results show that the main environmental factors affecting indoor PM concentration are the indoor heat index, indoor wind chill, wet bulb temperature and relative humidity. Comparing with Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) approaches, LATCN systematically reduced the prediction error rate (19.7% ~ 28.1% for the NAE, and 16.4% ~ 21.5% for the RMSE) and improved the model running speed (30.4% ~ 81.2%) over these classical sequence prediction models. Our study can inform the active prevention of indoor air pollution, and provides a theoretical basis for indoor environmental standards, while laying the foundations for developing novel air pollution prevention equipment in the future.

Potato Blight Detection Using Fine-Tuned CNN Architecture

Potato is one of the major cultivated crops and provides occupations and livelihoods for numerous people across the globe. It also contributes to the economic growth of developing and underdeveloped countries. However, potato blight is one of the major destroyers of potato crops worldwide. With the introduction of neural networks to agriculture, many researchers have contributed to the early detection of potato blight using various machine and deep learning algorithms. However, accuracy and computation time remain serious issues. Therefore, considering these challenges, we customised a convolutional neural network (CNN) to improve accuracy with fewer trainable parameters, less computation time, and reduced information loss. We compared the performance of the proposed model with various machine and deep learning algorithms used for potato blight classification. The proposed model outperformed the others with an overall accuracy of 99% using 839,203 trainable parameters in 183 s of training time.

Deep Ensemble-Based Approach Using Randomized Low-Rank Approximation for Sustainable Groundwater Level Prediction

Groundwater is the most abundant freshwater resource. Agriculture, industrialization, and domestic water supplies rely on it. The depletion of groundwater leads to drought. Topographic elevation, aquifer properties, and geomorphology influence groundwater quality. As the groundwater level data (GWL) are time series in nature, it is challenging to determine appropriate metrics and to evaluate groundwater levels accurately with less information loss. An effort has been made to forecast groundwater levels in India by developing a deep ensemble learning approach using a double-edge bi-directed long-short-term-memory (DEBi-LSTM) model approximated with a randomized low-ranked approximation algorithm (RLRA) and the variance inflation factor (VIF) to reduce information loss and to preserve data consistency. With minimal computation time, the model outperformed existing state-of-the-art models with 96.1% accuracy. To ensure sustainable groundwater development, the proposed work is discussed in terms of its managerial implications. By applying the model, we can identify safe, critical, and semi-critical groundwater levels in Indian states so that strategic plans can be developed.

Network-Based Characterization of Blood Large-Scale Coherent Motion in the Healthy Human Aorta With 4D Flow MRI.

The need for distilling the hemodynamic complexity of aortic flows into clinically relevant quantities resulted in a loss of the information hidden in 4D aortic fluid structures. To reduce information loss, this study proposes a network-based approach to identify and characterize in vivo the large-scale coherent motion of blood in the healthy human aorta. The quantitative paradigm of the aortic flow as a "social network" was applied on 4D flow MRI acquisitions performed on forty-one healthy volunteers. Correlations between the aortic blood flow rate waveform at the proximal ascending aorta (AAo), assumed as one of the drivers of aortic hemodynamics, and the waveforms of the axial velocity in the whole aorta were used to build "one-to-all" networks. The impact of the driving flow rate waveform and of aortic geometric attributes on the transport of large-scale coherent fluid structures was investigated. The anatomical length of persistence of large-scale coherent motion was the 29.6% of the healthy thoracic aorta length (median value, IQR 23.1%-33.9%). Such length is significantly influenced by the average and peak-to-peak AAo blood flow rate values, suggesting a remarkable inertial effect of the AAo flow rate on the transport of large-scale fluid structures in the distal aorta. Aortic geometric attributes such as curvature, torsion and arch shape did not influence the anatomical length of persistence. The proposed in vivo approach allowed to quantitatively characterize the transport of large-scale fluid structures in the healthy aorta, strengthening the definition of coherent hemodynamic structures and identifying flow inertia rather than geometry as one of its main determinants. The findings on healthy aortas may be used as reference values to investigate the impact of aortic disease or implanted devices in disrupting/restoring the physiological spatiotemporal coherence of large-scale aortic flow.

A progressive surrogate gradient learning for memristive spiking neural network

In recent years, spiking neural networks (SNNs) have received increasing attention of research in the field of artificial intelligence due to their high biological plausibility, low energy consumption, and abundant spatio-temporal information. However, the non-differential spike activity makes SNNs more difficult to train in supervised training. Most existing methods focusing on introducing an approximated derivative to replace it, while they are often based on static surrogate functions. In this paper, we propose a progressive surrogate gradient learning for backpropagation of SNNs, which is able to approximate the step function gradually and to reduce information loss. Furthermore, memristor cross arrays are used for speeding up calculation and reducing system energy consumption for their hardware advantage. The proposed algorithm is evaluated on both static and neuromorphic datasets using fully connected and convolutional network architecture, and the experimental results indicate that our approach has a high performance compared with previous research.

Safety Helmet Detection Based on YOLOv5 Driven by Super-Resolution Reconstruction.

High-resolution image transmission is required in safety helmet detection problems in the construction industry, which makes it difficult for existing image detection methods to achieve high-speed detection. To overcome this problem, a novel super-resolution (SR) reconstruction module is designed to improve the resolution of images before the detection module. In the super-resolution reconstruction module, the multichannel attention mechanism module is used to improve the breadth of feature capture. Furthermore, a novel CSP (Cross Stage Partial) module of YOLO (You Only Look Once) v5 is presented to reduce information loss and gradient confusion. Experiments are performed to validate the proposed algorithm. The PSNR (peak signal-to-noise ratio) of the proposed module is 29.420, and the SSIM (structural similarity) reaches 0.855. These results show that the proposed model works well for safety helmet detection in construction industries.

Feature Fusion Pyramid Network for End-to-end Scene Text Detection

How to properly involve text characteristics like multi-scale, arbitrary direction, length aspect ratio, into detection network design has become a hot topic in computer vision. Feature Pyramid Network (FPN) is a typical method to achieve robust text detection, where its low-level and high-level feature map retains spatial structure and global semantic information, respectively. However, its strict hierarchical structure fails to fuse low-level and high-level information to improve distinguish ability of feature map. To address this problem, we propose a novel feature fusion pyramid network for end-to-end scene text detection by fusing multi-modal information. By diving pyramid structure into high-level and low-level layers, channel and spatial attention modules are adopted to enhance high-level and low-level feature representation by encoding channel and spatial -wise context information, respectively. In order to reduce information loss by layer transmission, a special residual network is designed to achieve short-cut between high-level and low-level features, so as to realize multi-modal feature fusion. Experiments show the precision and recall of the propose method on ICDAR2015, ICDAR2017-MLT and MSRA-TD500 datasets reach 88.7%/82.1%, 77.0%/60.3% and 85.3%/74.8%, respectively.

Typhoon cloud image prediction based on enhanced multi-scale deep neural network

Typhoons threaten individuals’ lives and property. The accurate prediction of typhoon activity is crucial for reducing those threats and for risk assessment. Satellite images are widely used in typhoon research because of their wide coverage, timeliness, and relatively convenient acquisition. They are also important data sources for typhoon cloud image prediction. Studies on typhoon cloud image prediction have rarely used multi-scale features, which cause significant information loss and lead to fuzzy predictions with insufficient detail. Therefore, we developed an enhanced multi-scale deep neural network (EMSN) to predict a 3-hour-advance typhoon cloud image, which has two parts: a feature enhancement module and a feature encode-decode module. The inputs of the EMSN were eight consecutive images, and a feature enhancement module was applied to extract features from the historical inputs. To consider that the images of different time steps had different contributions to the output result, we used channel attention in this module to enhance important features. Because of the spatially correlated and spatially heterogeneous information at different scales, the feature encode-decode module used ConvLSTMs to capture spatiotemporal features at different scales. In addition, to reduce information loss during downsampling, skip connections were implemented to maintain more low-level information. To verify the effectiveness and applicability of our proposed EMSN, we compared various algorithms and explored the strengths and limitations of the model. The experimental results demonstrated that the EMSN efficiently and accurately predicted typhoon cloud images with higher quality than in the literature.

FE-YOLOv5: Feature enhancement network based on YOLOv5 for small object detection

Due to their inherent characteristics, small objects have weaker feature representation after multiple down-sampling and are even annihilated in the background. FPN’s simple feature concatenation does not fully utilize multi-scale information and introduces irrelevant context into the information transfer, further reducing the detection performance of the small object. To address the above issues, we propose the simple but effective FE-YOLOv5. (1) We designed the feature enhancement module (FEM) to capture more discriminative features of the small object. Global attention and high-level global contextual information are used to guide shallow, high-resolution features. Global attention interacts with cross-dimensional feature interaction and reduces information loss. High-level context complements more detailed semantic information by modeling global relationships through non-local networks. (2) We design the spatially aware module (SAM) to filter spatial information and enhance the robustness of features. Deformable convolution performs sparse sampling and adaptive spatial learning to better focus on foreground objects. According to the experimental results, our proposed FE-YOLOv5 outperforms the other architectures in the VisDrone2019 dataset and Tsinghua-Tencent100K dataset. Compared to YOLOv5, the APS was improved by 2.8% and 2.9%, respectively.

Hyperspectral Image Classification Using a Superpixel–Pixel–Subpixel Multilevel Network

Hyperspectral images (HSIs) often contain irregular ground cover with mixed spectral features and noise, which makes it challenging to identify the ground cover using only pixel features, superpixel features, or a combination of both. To alleviate the above problem, this paper proposes a superpixel-pixel-subpixel multilevel network (SPSM), which compensates for the insufficiencies of the different levels and decrease the information loss. For arbitrary irregular regions, superpixel features are simulated as network nodes using a graph convolutional network (GCN) to capture the spatial texture structure of the HSI, which improves the smooth classification results of local regions while facilitating the identification of different vegetation features in the region. Additionally, the global attention module (GAM) learns local regular regions based on pixel-level features to extend the global interactive representation capability and reduce information loss. To overcome spectral mixing and enhance material discrimination, the normalized attention module (NAM) is used to suppress unimportant subpixel information and identify and remove irrelevant details, thereby improving the identification of critical features that differentiate different materials. Finally, the three features are fused to build a SPSM classification framework to improve robustness to overfitting, reduce computational complexity, and facilitate target recognition. Experimental results on four HSI datasets demonstrate that the method is more capable of recognizing detailed features than other advanced comparison methods.