Enhancement Characteristics Research Articles

MSFA-Net: A Multiscale Feature Aggregation Network for Semantic Segmentation of Historical Building Point Clouds

In recent years, research on the preservation of historical architecture has gained significant attention, where the effectiveness of semantic segmentation is particularly crucial for subsequent repair, protection, and 3D reconstruction. Given the sparse and uneven nature of large-scale historical building point cloud scenes, most semantic segmentation methods opt to sample representative subsets of points, often leading to the loss of key features and insufficient segmentation accuracy of architectural components. Moreover, the geometric feature information at the junctions of components is cluttered and dense, resulting in poor edge segmentation. Based on this, this paper proposes a unique semantic segmentation network design called MSFA-Net. To obtain multiscale features and suppress irrelevant information, a double attention aggregation module is first introduced. Then, to enhance the model’s robustness and generalization capabilities, a contextual feature enhancement and edge interactive classifier module are proposed to train edge features and fuse the context data. Finally, to evaluate the performance of the proposed model, experiments were conducted on a self-curated ancient building dataset and the S3DIS dataset, achieving OA values of 95.2% and 88.7%, as well as mIoU values of 86.2% and 71.6%, respectively, further confirming the effectiveness and superiority of the proposed method.

Visible–Infrared Dual-Sensor Tracking Based on Transformer via Progressive Feature Enhancement and Fusion

Visible–Infrared Dual-Sensor Tracking Based on Transformer via Progressive Feature Enhancement and Fusion

Parallel Feature Enhancement and Adaptive Weighted Feature Fusion for Semantic Segmentation

Parallel Feature Enhancement and Adaptive Weighted Feature Fusion for Semantic Segmentation

A hybrid ICEEMDAN and OMEDA-based vibrodiagnosis method for the bearing of rolling stock

A hybrid ICEEMDAN and OMEDA-based vibrodiagnosis method for the bearing of rolling stock

Adaptive Scale and Correlative Attention PointPillars: An Efficient Real-Time 3D Point Cloud Object Detection Algorithm

Recognizing 3D objects from point clouds is a crucial technology for autonomous vehicles. Nevertheless, LiDAR (Light Detection and Ranging) point clouds are generally sparse, and they provide limited contextual information, resulting in unsatisfactory recognition performance for distant or small objects. Consequently, this article proposes an object recognition algorithm named Adaptive Scale and Correlative Attention PointPillars (ASCA-PointPillars) to address this problem. Firstly, an innovative adaptive scale pillars (ASP) encoding method is proposed, which encodes point clouds using pillars of varying sizes. Secondly, ASCA-PointPillars introduces a feature enhancement mechanism called correlative point attention (CPA) to enhance the feature associations within each pillar. Additionally, a data augmentation algorithm called random sampling data augmentation (RS-Aug) is proposed to solve the class imbalance problem. The experimental results on the KITTI 3D object dataset demonstrate that the proposed ASCA-PointPillars algorithm significantly boosts the recognition performance and RS-Aug effectively enhances the training effects on an imbalanced dataset.

Improved hypertensive stroke classification based on multi-scale feature fusion of head axial CT angiogram and multimodal learning

Purpose:Strokes are severe cardiovascular and circulatory diseases with two main types: ischemic and hemorrhagic. Clinically, brain images such as computed tomography (CT) and computed tomography angiography (CTA) are widely used to recognize stroke types. However, few studies have combined imaging and clinical data to classify stroke or consider a factor as an Independent etiology. Methods:In this work, we propose a classification model that automatically distinguishes stroke types with hypertension as an independent etiology based on brain imaging and clinical data. We first present a preprocessing workflow for head axial CT angiograms, including noise reduction and feature enhancement of the images, followed by an extraction of regions of interest. Next, we develop a multi-scale feature fusion model that combines the location information of position features and the semantic information of deep features. Furthermore, we integrate brain imaging with clinical information through a multimodal learning model to achieve more reliable results. Results:Experimental results show our proposed models outperform state-of-the-art models on real imaging and clinical data, which reveals the potential of multimodal learning in brain disease diagnosis. Conclusion:The proposed methodologies can be extended to create AI-driven diagnostic assistance technology for categorizing strokes.

FEMSFNet: Feature Enhancement and Multi-Scales Fusion Network for SAR Aircraft Detection

Aircraft targets, as high-value subjects, are a focal point in Synthetic Aperture Radar (SAR) image interpretation. To tackle challenges like limited SAR aircraft datasets and shortcomings in existing detection algorithms (complexity, poor performance, weak generalization), we present the Feature Enhancement and Multi-Scales Fusion Network (FEMSFNet) for SAR aircraft detection. FEMSFNet employs diverse image augmentation and integrates optimized Squeeze-and-Excitation Networks (SE) with residual network (ResNet) in a SdE-Resblock structure for a lightweight yet accurate model. It introduces ssppf-CSP module, an improved pyramid pooling model, to prevent receptive field deviation in deep network training. Tailored for SAR aircraft detection, FEMSFNet optimizes loss functions, emphasizing both speed and accuracy. Evaluation on the SAR Aircraft Detection Dataset (SADD) demonstrates significant improvements compared to the contrasted algorithms: precision rate (92%), recall rate (96%), and F1 score (94%), with a maximum increase of 12.2% in precision, 12.9% in recall, and 13.3% in F1 score.

A novel defect prediction method based on semantic feature enhancement

SummaryAlthough cross‐project defect prediction (CPDP) techniques that use traditional manual features to build defect prediction model have been well‐developed, they usually ignore the semantic and structural information inside the program and fail to capture the hidden features that are critical for program category prediction, resulting in poor defect prediction results. Researchers have proposed using deep learning to automatically extract the semantic features of programs and fuse them with traditional features as training data. However, in practice, it is important to explore the effective representation of the semantic features in the programs and how the fusion of a reasonable ratio between the two types of features can maximize the effectiveness of the model. In this paper, we propose a semantic feature enhancement‐based defect prediction framework (SFE‐DP), which augments the semantic feature set extracted from the program code with data. We also introduce a layer of self‐attentive mechanism and a matching layer to filter low‐efficiency and non‐critical semantic features in the model structure. Finally, we combine the idea of hybrid loss function to iteratively optimize the model parameters. Extensive experiments validate that SFE‐DP can outperform the baseline approaches on 90 pairs of CPDP tasks formed by 10 open‐source projects.

Context Aggregation Network for Remote Sensing Image Semantic Segmentation

In recent years, remote sensing technology has been widely applied in various industries, and semantic segmentation of remote sensing images has attracted much attention. Due to the complexity and special characteristics of remote sensing images, multi-scale object detection and accurate object localization are important challenges in remote sensing image semantic segmentation. Therefore, this paper proposes a context aggregation network (CANet). The design of CANet is influenced by advanced technologies such as attention mechanisms and feature fusion and enhancement. This network first introduces nested dilated residual module (NDRM), which can fully utilize the features extracted by the backbone network. Then, improved integrated successive dilation module (IISD) is proposed to effectively aggregate a series of contextual information scales. Next, Swim Transformer module is embedded to provide global contextual information. Finally, multi-resolution fusion module (MRFM) is proposed, allowing the comprehensive fusion of feature layers from different stages of the encoder, preserving more semantic and detailed information. The experimental results show that CANet outperforms other advanced models on the Potsdam and Vaihingen datasets.

Accuracy and efficiency stereo matching network with adaptive feature modulation.

Feature enhancement plays a crucial role in improving the quality and discriminative power of features used in matching tasks. By enhancing the informative and invariant aspects of features, the matching process becomes more robust and reliable, enabling accurate predictions even in challenging scenarios, such as occlusion and reflection in stereo matching. In this paper, we propose an end-to-end dual-dimension feature modulation network called DFMNet to address the issue of mismatches in interference areas. DFMNet utilizes dual-dimension feature modulation (DFM) to capture spatial and channel information separately. This approach enables the adaptive combination of local features with more extensive contextual information, resulting in an enhanced feature representation that is more effective in dealing with challenging scenarios. Additionally, we introduce the concept of cost filter volume (CFV) by utilizing guide weights derived from group-wise correlation. CFV aids in filtering the concatenated volume adaptively, effectively discarding redundant information, and further improving matching accuracy. To enable real-time performance, we designed a fast version named Fast-GFM. Fast-GFM employs the global feature modulation (GFM) block to enhance the feature expression ability, improving the accuracy and stereo matching robustness. The accurate DFMNet and the real-time Fast-GFM achieve state-of-the-art performance across multiple benchmarks, including Scene Flow, KITTI, ETH3D, and Middlebury. These results demonstrate the effectiveness of our proposed methods in enhancing feature representation and significantly improving matching accuracy in various stereo matching scenarios.

[Automatic epilepsy detection with an attention-based multiscale residual network].

The deep learning-based automatic detection of epilepsy electroencephalogram (EEG), which can avoid the artificial influence, has attracted much attention, and its effectiveness mainly depends on the deep neural network model. In this paper, an attention-based multi-scale residual network (AMSRN) was proposed in consideration of the multiscale, spatio-temporal characteristics of epilepsy EEG and the information flow among channels, and it was combined with multiscale principal component analysis (MSPCA) to realize the automatic epilepsy detection. Firstly, MSPCA was used for noise reduction and feature enhancement of original epilepsy EEG. Then, we designed the structure and parameters of AMSRN. Among them, the attention module (AM), multiscale convolutional module (MCM), spatio-temporal feature extraction module (STFEM) and classification module (CM) were applied successively to signal reexpression with attention weighted mechanism as well as extraction, fusion and classification for multiscale and spatio-temporal features. Based on the Children's Hospital Boston-Massachusetts Institute of Technology (CHB-MIT) public dataset, the AMSRN model achieved good results in sensitivity (98.56%), F1 score (98.35%), accuracy (98.41%) and precision (98.43%). The results show that AMSRN can make good use of brain network information flow caused by seizures to enhance the difference among channels, and effectively capture the multiscale and spatio-temporal features of EEG to improve the performance of epilepsy detection.

MSGAT: Multi-scale gated axial reverse attention transformer network for medical image segmentation

Medical image segmentation can help doctors accurately identify and locate various structures, tissues, or lesions in the patient's body, providing an important basis for clinical diagnosis. However, the segmentation of medical images is seriously affected by the complexity of the structures in the images, the variety of morphologies, the uneven contrast, and the blurring of the boundaries between the target tissues and the background. Therefore, we propose a Multiscale Gated Axial Reverse Attention Transformer (MSGAT) for medical image segmentation. First, we use the Pyramid Vision Transformer (PVT) as the backbone network to extract the high-level features and introduce the Channel Received Field Block (CRFB) to filter the interfering information in the high-level features, and then we augment and aggregate the high-level features separately. Specifically, on the one hand, the Multiscale Feature Enhancement Module (MSFEM) is used to obtain multiscale contextual information to enhance the positional information of the target, and on the other hand, Multiscale Parallel Partial Decoder (MSPPD) is employed to aggregate the high-level feature maps to generate rough segmentation maps. Finally, the rough segmentation map is used to guide the gated axial reverse attention (GARA) combined with the position information in the enhanced feature map to refine the segmentation boundaries and gradually mine the detailed information of the target. MSGAT is evaluated against state-of-the-art methods on skin cancer (ISIC2016, PH2), polyp (Kvair), and gland (GlaS) datasets. Results demonstrate MSGAT's superiority, with an average improvement of 6.19% and 9.35% in mIoU and DSC metrics, respectively, compared to existing methods.

A single-frame infrared small target detection method based on joint feature guidance

AbstractSingle-frame infrared small target detection is affected by the low image resolution and small target size, and is prone to the problems of small target feature loss and positional offset during continuous downsampling; at the same time, the sparse features of the small targets do not correlate well with the global-local linkage of the background features. To solve the above problems, this paper proposes an efficient infrared small target detection method. First, this paper incorporates BlurPool in the feature extraction part, which reduces the loss and positional offset of small target features in the process of convolution and pooling. Second, this paper designs an interactive attention deep feature fusion module, which acquires the correlation information between the target and the background from a global perspective, and designs a compression mechanism based on deep a priori knowledge, which reduces the computational difficulty of the self-attention mechanism. Then, this paper designs the context local feature enhancement and fusion module, which uses deep semantic features to dynamically guide shallow local features to realize enhancement and fusion. Finally, this paper proposes an edge feature extraction module for shallow features, which utilizes the complete texture and location information in the shallow features to assist the network to initially locate the target position and edge shape. Numerous experiments show that the method in this paper significantly improves nIoU, F1-Measure and AUC on IRSTD-1k Datasets and NUAA-SIRST Datasets.

Bi-directional information guidance network for UAV vehicle detection

AbstractUAV vehicle detection based on convolutional neural network exits a key problem of information imbalance of different feature layers. Shallow features have spatial information that is beneficial to localization, but lack semantic information. On the contrary, deep features have semantic information that is beneficial to classification, but lack spatial information. However, accurate classification and localization for UAV vehicle detection require both shallow spatial information and high semantic information. In our work, a bi-directional information guidance network (BDIG-Net) for UAV vehicle detection is proposed, which can ensure that each feature prediction layer has abundant mid-/low-level spatial information and high-level semantic information. There are two main parts in the BDIG-Net: shallow-level spatial information guidance part and deep-level semantic information guidance part. In the shallow-level guidance part, we design a feature transform module (FTM) to supply the mid-/low-level feature information, which can guide the BDIG-Net to enhance detailed and spatial features for deep features. Furthermore, we adopt a light-weight attention module (LAM) to reduce unnecessary shallow background information, making the network more focused on small-sized vehicles. In the deep-level guidance part, we use classical feature pyramid network to supply high-level semantic information, which can guide the BDIG-Net to enhance contextual information for shallow features. Meanwhile, we design a feature enhancement module (FEM) to suppress redundant features and improve the discriminability of vehicles. The proposed BDIG-Net can reduce the information imbalance. The experimental results show that the BDIG-Net can achieve accurate classification and localization for UAV vehicles and realize the real-time application requirements.

A large-scale point cloud semantic segmentation neural network based on long-range contextual dependencies enhancement

ABSTRACT This letter presents a long-range contextual dependency enhanced network (LCDE-Net) for semantic segmentation of large-scale point cloud, which employs a U-shaped framework. Firstly, point clouds are subsampled with grid sampling and fed into convolutional layers to learn more representative local features of points. Then global and local encoders (GLE) are designed to exploit long-range contextual dependencies and local features simultaneously. The core of GLE consists of two parts: global feature enhancement (GFE) module and feature channel modulation (FCM) module. Secondly, through decoder layers, the encoded features are upsampled through the nearest-neighbour interpolation and aggregated with the intermediate encoded features by skip connections to capture multi-scale discriminative features for semantic segmentation of point cloud. Finally, via Fully Connection layer and Softmax classifier, each point’s label is assigned. Two different benchmark datasets are conducted to evaluate the performance of the proposed method, Experimental results report that the proposed LCDE-Net achieves 78.6% in terms of mean intersection over union (mIoU) on Semantic3D, and 68.2% on S3DIS, which is the highest among the comparison methods. The code of LCDE-Net is available at https://github.com/xrzmyz/LCDE-Net.

D3Former: Jointly learning repeatable dense detectors and feature-enhanced descriptors via saliency-guided transformer

Establishing accurate and representative matches is a crucial step in addressing the point cloud registration problem. A commonly employed approach involves detecting keypoints with salient geometric features and subsequently mapping these keypoints from one frame of the point cloud to another. However, methods within this category are hampered by the repeatability of the sampled keypoints. In this paper, we introduce a saliency-guided transformer, referred to as D3Former, which entails the joint learning of repeatable Dense Detectors and feature-enhanced Descriptors. The model comprises a Feature Enhancement Descriptor Learning (FEDL) module and a Repetitive Keypoints Detector Learning (RKDL) module. The FEDL module utilizes a region attention mechanism to enhance feature distinctiveness, while the RKDL module focuses on detecting repeatable keypoints to enhance matching capabilities. Extensive experimental results on challenging indoor and outdoor benchmarks demonstrate that our proposed method consistently outperforms state-of-the-art point cloud matching methods. Notably, tests on 3DLoMatch, even with a low overlap ratio, show that our method consistently outperforms recently published approaches such as RoReg and RoITr. For instance, with the number of extracted keypoints reduced to 250, the registration recall scores for RoReg, RoITr, and our method are 64.3%, 73.6%, and 76.5%, respectively.

On-machine measurement and compensation of thin-walled surface

On-machine measurement technology is considered as a key technology for realizing closed-loop feedback control in intelligent manufacturing due to the reduction of the transfer process. However, the complexity of the machine tool process system introduces some uncertainty into the accuracy of on-machine measurement, which severely limits the application in the actual industrial field. To overcome the shortcomings of the existing uncertainty, an inspection framework for on-machine measurement of thin-walled surface is proposed. Firstly, a low-cost wireless on-machine measurement system based on potential signals is established and integrated into a manufacturing process line for automatic sampling. Then, the similarity of momentum conservation is introduced into sampling planning, and an adaptive sampling model based on momentum conservation and multi-objective particle swarm optimizer is proposed. Finally, a stacked deep learning model under the vertical inspection direction is proposed to improve the inspection accuracy by correcting the sampling data. Compared with existing sampling methods, the proposed model is similar to an attention mechanism that enables adaptive enhancement of profile features. The inspection performance by data correction is improved by about 16.14% in the mean inspection error. Simulations and experiments show that the proposed method has great advantages in terms of efficiency and robustness, which can provide a theoretical reference for adaptive toolpath correction for thin-walled surfaces.

Recursive attention collaboration network for single image de‐raining

AbstractSingle‐image rain removal is an important problem in the field of computer vision aimed at recovering clean images from rainy images. In recent years, data‐driven convolutional neural network (CNN)‐based rain removal methods have achieved significant results, but most of them cannot fully focus on the contextual information in rain‐containing images, which leads to the failure of recovering some of the background details of the images that have been corrupted due to the aggregation of rain streaks. With the success of Transformer‐based models in the field of computer vision, global features can be easily acquired to better help recover details in the background of an image. However, Transformer‐based models often require a large number of parameters during the training process, which makes the training process very difficult and makes it difficult to apply them to specific devices for execution in reality. The authors propose a Recursive Attention Collaboration Network, which consists of a recursive Swin‐transformer block (STB) and a CNN‐based feature fusion block. The authors designed the Recursively Integrate Transformer Block (RITB), which consists of several STBs recursively connected, that can effectively reduce the number of parameters of the model. The final part of the module can integrate the local information from the STBs. The authors also design the Feature Enhancement Block, which can better recover the details of the background information corrupted by rain streaks of different density shapes through the features passed from the RITB. Experiments show that the proposed network has an effective rain removal effect on both synthetic and real datasets and has fewer model parameters than other mainstream methods.

An acoustic weighing method based on oscillation signal and feature enhanced network

Acoustic weighing is a promising method for non-contact mass measurement of tiny objects as it avoids contamination and contact losses. However, due to the highly nonlinear nature of the acoustic field, some parameters of the mechanism model of acoustic weighing cannot be accurately simulated, thereby reducing the accuracy of acoustic weighing. To improve the accuracy of acoustic weighing, we propose an acoustic weighing method based on oscillating signals and feature enhancement network. Firstly, to drive the object oscillation and collect oscillation data, an acoustic levitation-based data acquisition system is constructed. Then, to break the limitations of the mechanism model, a feature enhancement network named CNN-BiLSTM-SE is proposed, which directly establishes the correlation between oscillating signals and actual mass. Finally, these data are used to train and test the proposed network model, validating the effectiveness of the model. Experimental results show that the method achieves high accuracy in measuring object mass, following the actual measurements with remarkable consistency. In addition, our approach is also suitable for acoustic weighing of small and sensitive objects, opening up new perspective for the study and application of nonlinear acoustic systems.

PTDS CenterTrack: pedestrian tracking in dense scenes with re-identification and feature enhancement

PTDS CenterTrack: pedestrian tracking in dense scenes with re-identification and feature enhancement