Multiple Feature Fusion Research Articles

PE-MCAT: Leveraging Image Sensor Fusion and Adaptive Thresholds for Semi-Supervised 3D Object Detection.

Existing 3D object detection frameworks in sensor-based applications heavily rely on large-scale annotated data to achieve optimal performance. However, obtaining such annotations from sensor data-like LiDAR or image sensors-is both time-consuming and costly. Semi-supervised learning offers an efficient solution to this challenge and holds significant potential for sensor-driven artificial intelligence (AI) applications. While it reduces the need for labeled data, semi-supervised learning still depends on a small amount of labeled samples for training. In the initial stages, relying on such limited samples can adversely affect the effective training of student-teacher networks. In this paper, we propose PE-MCAT, a semi-supervised 3D object detection method that generates high-precision pseudo-labels. First, to address the challenges of insufficient local feature capture and poor robustness in point cloud data, we introduce a point enrichment module. This module incorporates information from image sensors and combines multiple feature fusion methods of local and self-features to directly enhance the quality of point clouds and pseudo-labels, compensating for the limitations posed by using only a few labeled samples. Second, we explore the relationship between the teacher network and the pseudo-labels it generates. We propose a multi-class adaptive threshold strategy to initially filter and create a high-quality pseudo-label set. Furthermore, a joint variable threshold strategy is introduced to refine this set further, enhancing the selection of superior pseudo-labels.Extensive experiments demonstrate that PE-MCAT consistently outperforms recent state-of-the-art methods across different datasets. Specifically, on the KITTI dataset and using only 2% of labeled samples, our method improved the mean Average Precision (mAP) by 0.7% for cars, 3.7% for pedestrians, and 3.0% for cyclists.

Medium- and Long-Term Distributed Photovoltaic Power Prediction Based on Multiple Time Series Feature and Multiple-Model Fusion

Distributed photovoltaic power stations have advantages such as local direct power supply and reduced transmission energy consumption, and whose demands are constantly being developed. Conducting research on medium- and long-term distributed photovoltaic prediction will have significant value for applications such as the electricity trade market, power grid operation, and the planning of new power stations. Due to characteristics such as long time dependence, disperse power stations, and strong randomness, making accurate and stable predictions becomes very difficult. In this research, we propose a multiple time series feature and multiple-model fusion-based ensemble learning model for medium- and long-term distributed photovoltaic power prediction (M2E-DPV). Considering the wave influence and the differences in distributions in different areas of photovoltaic power, multiple feature combinations are designed to increase feature expression ability and adaptability. Based on the boost ensemble learning model, trained on a single model of different time scale features, the optimal scoring strategy is used for multiple model fusion in the rolling prediction process, and finally, time-segmented probabilistic correction is performed. The experiment results show the effectiveness of the M2E-DPV under multiple feature combinations and multi-model fusion strategies. The average MAPE, R2, and ACC indicators are 0.15, 0.96, and 0.91, respectively. Compared with other methods, there is a significant improvement, indicating that the prediction ability of the model framework proposed in this paper is advanced and robust.

3D Point Cloud Semantic Segmentation based on Multi-scale Dense Nested Networks

Aiming at the problem that the relationship between geometric features and semantic features is ignored in the point cloud data downsampling process, which leads to inaccurate segmentation of object boundaries and structural details, this paper proposes a 3D point cloud semantic segmentation network based on multi-scale dense nested type. Firstly, a dense nested network architecture is constructed by nesting multiple multi-scale feature fusion modules to fuse multi-scale features of different directions between encoder-decoder paths, so as to effectively propagate local ge-ometric context information and enhance the ability of cross-scale information interaction. Secondly, a local feature aggregation unit is constructed in the multi-scale feature fusion module, which strengthens the structural awareness within the local point set based on graph convolution and attention mechanism, and promotes the complementarity of local geometric features and abstract semantic information. Then, the cross-layer multi-loss supervision module is combined to further optimize the multi-scale feature propagation, which makes the network training more stable and improves the point cloud segmentation accuracy. Finally, this paper verified the proposed network on the S3DIS dataset. The experimental results show that the proposed network has the mean intersection over Union of 71.2% and the overall accuracy of 88.7%, which is 1.2 and 0.7 percentage points higher than RandLA-Net, respectively, which proves that the proposed network can effectively improve the accuracy of 3D point cloud semantic segmentation.

Open set identification of malicious encrypted traffic based on multi-feature fusion

In the current network environment, an increasing amount of malicious traffic is transmitted through encrypted channels, carrying control commands and data. With the continuous development of communication protocols and applications, new types of malicious encrypted traffic are emerging, posing significant challenges for network management (e.g., traffic engineering). Therefore, accurately identifying malicious traffic in complex open network spaces has become a hot research topic in network security. In this study, we draw inspiration from channel theory in image science and innovatively convert traffic data into Red-Green-Blue (RGB) image format to achieve the fusion of multiple features. Inspired by image recognition technologies, we have designed a multi-granularity network model that integrates both global and local features, serving as our core network architecture. At the top of the model, we have equipped each known category with a unique autoencoder, using its generated manifold to replace traditional prototypes for model construction. Classification is accomplished through a scoring mechanism that evaluates category membership and by setting thresholds to achieve open set recognition of unknown categories. Relying on our self-created dataset,Malicious and Encrypted Traffic 2024 (MNET2024), we conduct a series of extensive experiments. The results demonstrate that our proposed method exhibits outstanding performance in both closed-set and open-set recognition tasks.

A Chinese Nested Named Entity Recognition Model for Chicken Disease Based on Multiple Fine-Grained Feature Fusion and Efficient Global Pointer

Extracting entities from large volumes of chicken epidemic texts is crucial for knowledge sharing, integration, and application. However, named entity recognition (NER) encounters significant challenges in this domain, particularly due to the prevalence of nested entities and domain-specific named entities, coupled with a scarcity of labeled data. To address these challenges, we compiled a corpus from 50 books on chicken diseases, covering 28 different disease types. Utilizing this corpus, we constructed the CDNER dataset and developed a nested NER model, MFGFF-BiLSTM-EGP. This model integrates the multiple fine-grained feature fusion (MFGFF) module with a BiLSTM neural network and employs an efficient global pointer (EGP) to predict the entity location encoding. In the MFGFF module, we designed three encoders: the character encoder, word encoder, and sentence encoder. This design effectively captured fine-grained features and improved the recognition accuracy of nested entities. Experimental results showed that the model performed robustly, with F1 scores of 91.98%, 73.32%, and 82.54% on the CDNER, CMeEE V2, and CLUENER datasets, respectively, outperforming other commonly used NER models. Specifically, on the CDNER dataset, the model achieved an F1 score of 79.68% for nested entity recognition. This research not only advances the development of a knowledge graph and intelligent question-answering system for chicken diseases, but also provides a viable solution for extracting disease information that can be applied to other livestock species.

Fast Semantic Segmentation of Ultra-High-Resolution Remote Sensing Images via Score Map and Fast Transformer-Based Fusion

For ultra-high-resolution (UHR) image semantic segmentation, striking a balance between computational efficiency and storage space is a crucial research direction. This paper proposes a Feature Fusion Network (EFFNet) to improve UHR image semantic segmentation performance. EFFNet designs a score map that can be embedded into the network for training purposes, enabling the selection of the most valuable features to reduce storage consumption, accelerate speed, and enhance accuracy. In the fusion stage, we improve upon previous redundant multiple feature fusion methods by utilizing a transformer structure for one-time fusion. Additionally, our combination of the transformer structure and multibranch structure allows it to be employed for feature fusion, significantly improving accuracy while ensuring calculations remain within an acceptable range. We evaluated EFFNet on the ISPRS two-dimensional semantic labeling Vaihingen and Potsdam datasets, demonstrating that its architecture offers an exceptionally effective solution with outstanding semantic segmentation precision and optimized inference speed. EFFNet substantially enhances critical performance metrics such as Intersection over Union (IoU), overall accuracy, and F1-score, highlighting its superiority as an architectural innovation in ultra-high-resolution remote sensing image semantic segmentation.

Failure Feature Identification of Vibrating Screen Bolts under Multiple Feature Fusion and Optimization Method

Failure Feature Identification of Vibrating Screen Bolts under Multiple Feature Fusion and Optimization Method

Prediction of Super-enhancers Based on Mean-shift Undersampling

Background: Super-enhancers are clusters of enhancers defined based on the binding occupancy of master transcription factors, chromatin regulators, or chromatin marks. It has been reported that super-enhancers are transcriptionally more active and cell-type-specific than regular enhancers. Therefore, it is necessary to identify super-enhancers from regular enhancers. A variety of computational methods have been proposed to identify super-enhancers as auxiliary tools. However, most methods use ChIP-seq data, and the lack of this part of the data will make the predictor unable to execute or fail to achieve satisfactory performance. Objective: The aim of this study is to propose a stacking computational model based on the fusion of multiple features to identify super-enhancers in both human and mouse species. Methods: This work adopted mean-shift to cluster majority class samples and selected four sets of balanced datasets for mouse and three sets of balanced datasets for human to train the stacking model. Five types of sequence information are used as input to the XGBoost classifier, and the average value of the probability outputs from each classifier is designed as the final classification result. Results: The results of 10-fold cross-validation and cross-cell-line validation prove that our method has superior performance compared to other existing methods. The source code and datasets are available at https://github.com/Cheng-Han-max/SE_voting. Conclusion: The analysis of feature importance indicates that Mismatch accounts for the highest proportion among the top 20 important features.

Prediction of hypertension risk based on multiple feature fusion

ObjectiveIn the application of machine learning to the prediction of hypertension, many factors have seriously affected the classification accuracy and generalization performance. We propose a pulse wave classification model based on multi-feature fusion for accuracy prediction of hypertension. Methods and MaterialsWe propose an ensemble under-sampling model with dynamic weights to decrease the influence of class imbalance on classification, further to automatically classify of hypertension on inquiry diagnosis. We also build a deep learning model based on hybrid attention mechanism, which transforms pulse waves to feature maps for extraction of in-depth features, so as to automatically classify hypertension on pulse diagnosis. We build the multi-feature fusion model based on dynamic Dempster/Shafer (DS) theory combining inquiry diagnosis and pulse diagnosis to enhance fault tolerance of prediction for multiple classifiers. In addition, this study calculates feature importance ranking of scale features on inquiry diagnosis and temporal and frequency-domain features on pulse diagnosis. ResultsThe accuracy, sensitivity, specificity, F1-score and G-mean after 5-fold cross-validation were 94.08%, 93.43%, 96.86%, 93.45% and 95.12%, respectively, based on the hypertensive samples of 409 cases from Longhua Hospital affiliated to Shanghai University of Traditional Chinese Medicine and Hospital of Integrated Traditional Chinese and Western Medicine. We find the key factors influencing hypertensive classification accuracy, so as to assist in the prevention and clinical diagnosis of hypertension. ConclusionCompared with the state-of-the-art models, the multi-feature fusion model effectively utilizes the patients’ correlated multimodal features, and has higher classification accuracy and generalization performance.

Fusion of Multiple Basic Element Features for Airborne LiDAR in-house Surveys

When using airborne LiDAR point clouds for city modelling and road extraction, point cloud classification is a crucial step. There are numerous ways for classifying point clouds, but there are still issues like redundant multi-dimensional feature vector data and poor point cloud classification in intricate situations. A point cloud classification method built on the fusing of multikernel feature vectors is suggested as a solution to these issues. The technique employs random forest to classify point cloud data by merging colour information, and it extracts feature vectors based on point primitives and object primitives, respectively. In this study, a densely populated area was chosen as the study area. Light airborne LIDAR mounted on a delta wing was used to collect point cloud data at a low altitude (170 m) over a dense cross-course. The point cloud data were then combined, corrected, and enhanced with texture data, and the houses were vectorized on the point cloud. The accuracy of the results was then assessed. With a median inaccuracy of 4.8 cm and a point cloud data collection rate of 83.3%, using airborne LIDAR to measure house corners can significantly lighten the labour associated with external house corner measurements.This test extracts the texture information of point cloud data through the efficient processing of high-density point cloud data, providing a reference for the application of texture information of airborne LIDAR data and a clear understanding of its accuracy.

Incomplete Footprint Retrieval Based on Multi-Scale Feature Orthogonal Fusion

At present, footprint image retrieval based on deep learning mainly focuses on complete footprint, but many of the footprints obtained in the field of public safety and criminal investigation are incomplete forms, Therefore, the feature analysis of incomplete footprint has important practical significance. Based on multiple scale features fusion, we proposed a method to solve incomplete footprints retrieval. On the basis of extracting the global feature of footprint, this method simultaneously extracts multiple local features from different stages of the backbone network to supplement the footprint feature information. The multi-scale feature orthogonal fusion module is used to reduce redundant features, improve the expression ability of footprint features, and solve the problem of incomplete footprint retrieval to a certain extent. The experiment shows that our method has certain effectiveness in retrieving problems on incomplete footprint, with a Top 1 accuracy of 87.08%, expanding the scope of footprint research.

Map-based localization for intelligent vehicles based on fusion of multiple visual features in underground parking

The ‘last mile’ problem in intelligent car localization within underground parking lots presents challenges such as localization blind spots and feature extraction difficulties. To tackle these issues, this paper proposes a mapping and multiscale matching localization method based on the fusion of multiple feature sources using a ‘binocular mapping, monocular localization’ approach. By combining traditional features with deep learning features, the proposed method enhances feature matching accuracy and addresses feature extraction challenges. A sparse three-dimensional reconstruction method based on local feature-optical flow is also proposed to mitigate localization blind spots. During the mapping process, the underground parking lot scene is segmented into several nodes to store map information, each consisting of a feature layer, scene structure layer, and trajectory layer. In the multiscale matching localization process, ORB, LoFTR, and BoW multisource features are fused to achieve feature matching, determine the nearest node, and estimate vehicle pose using the PnP model through optical flow. Finally, a localization optimization algorithm outputs the localization result. Experiments were conducted across multiple underground parking lots encompassing a total area of over 100,000 square meters. The results demonstrated that the average error of map construction using the proposed algorithm is approximately 15 mm, and the average error rate of localization is 4.45%.

Anti-Occlusion Visual Tracking Algorithm for UAVs with Multi-Feature Adaptive Fusion

Most of the existing trackers based on discriminative correlation filters use only one feature or a simple linear fusion of multiple features for object tracking, and most of them lack a mechanism to handle occlusions. This leads to poor tracking performance in rapidly changing and easily occluded scenarios, especially on unmanned aerial vehicle (UAV) platforms. To address this issue, this paper proposes an anti-occlusion visual tracking algorithm for UAVs with multi-feature adaptive fusion named multi-feature adaptive fusion and anti-occlusion tracker (MAFAOT). It introduces a novel approach for implementing an adaptive fusion of multiple features. This method transforms the multi-feature fusion problem into a maximization issue by designing a tracking quality evaluation index. It successfully achieves an adaptive fusion of gradient direction histogram and color histogram feature responses. MAFAOT also introduces an anti-occlusion update pool strategy, enabling the tracker to adapt dynamically to various complex scenarios, including occlusion and motion blur. The experimental results on the OTB100 and UAV123 datasets confirm the significant advantages of MAFAOT in terms of precision and success rate compared to other correlation filter-based algorithms. The proposed methods further enhance the expressiveness of the features and effectively avoid the problem of tracker contamination caused by occlusion. Furthermore, this paper applies the proposed methods to the kernelized correlation filters (KCF) algorithm. On the OTB100 dataset, the improved KCF algorithm shows an improvement of 10.94% in precision and 11.11% in success rate. On the UAV123 dataset, it shows an improvement of 14.53% in precision and 16.62% in success rate, further verifying the effectiveness and versatility of the proposed methods.

A multiple feature fusion-based intelligent optimization ensemble model for carbon price forecasting

Accurately predicting carbon emissions trading price can assist governments in devising policies that strike a balance between economic development and environmental protection. In recent years, the decomposition integration framework has been widely applied for carbon price prediction. However, due to the inadequacy of feature extraction and lack of emphasis on interval forecasting, constructing a stable and efficient forecasting model for carbon price remains a challenging task. Thus, this study proposes an enhanced decomposition integration model for carbon price prediction that conducts multiple feature fusion based on the potential determinants and establishes a multivariate intelligent optimized ensemble model. Firstly, decomposition and entropy algorithms are employed to decompose and reconstruct the carbon price into several simple subsequences to strike the balance between computing speed and the complexity of the model. Subsequently, considering how external factors may affect the sequence, structured data, and unstructured data are both introduced to construct the multiple forecasting model. An intelligent optimization ensemble model is then developed for predicting the reconstructed components based on a deep learning model with an attention mechanism. Lastly, to quantify uncertainty, interval prediction is incorporated for research and analysis. Empirical analysis is conducted on three carbon finance pilot markets in China and the values of Mean Absolute Percentage Error (MAPE) are 2.047, 1.251, and 0.994 respectively. Furthermore, each module of the proposed model significantly contributes to the results and the developed model outperforms the baseline model across various error metrics.

Infrared small target detection algorithm based on filter kernel combination optimization learning method

Infrared small target detection in complex backgrounds remains a big challenge due to its small size and weak energy, which makes it easy to be submerged in complex backgrounds. Common filtering algorithms are difficult to apply to various complex scenes and generate many false alarms and poor robustness. Therefore, this paper proposes an infrared small target detection algorithm based on filter kernel combination optimization learning method. For the sample image block use typical filter kernels and weight coefficients to calculate the combination of filtering results, and construct the loss function of the results and the true value. The optimal weight coefficients are learned using an optimization method, which effectively achieves the fusion of multiple features, then we build a sample library containing comprehensive priori information. In the implementation, an online inference method based on similarity measure is proposed to match test images with sample library image blocks to obtain optimal combination weights. By utilizing the comprehensive priori information to enhance the adaptive ability of the algorithm for various complex backgrounds. Finally, get the position of the infrared small target. The experimental results show that our algorithm has excellent detection performance for infrared small targets with complex background and low signal to clutter ratio, and the AUC value is 0.9980, which is more robust than other comparative algorithms.

Intelligent fault diagnosis of roller bearing based on dual-flow convolutional neural network

Deep learning (DL) has inaugurated new approaches to implement fault diagnosis of roller bearings, which is essential to deal with the current industrial big data era. Unfortunately, many existing deep learning models, particularly convolutional neural network (CNN) models, have the following drawbacks. Single-channel convolution neural network in pooling layer has the problem of data loss. The feature information extracted by CNN is not integrated due to lack of feature learning ability, which will induce unsatisfactory diagnostic results and poor generalization ability. To deal with the above issues, a dual-flow convolutional neural network (DFCNN) is proposed for intelligent diagnosis the faults of roller bearings. The multi-channel convolutional neural network is used for feature extraction to solve the problem of data loss in the pooling layer. More comprehensive fault features can be extracted by multiple feature fusion after the input of two dimensions. Network parameters are adjusted using model optimization. The proposed method is demonstrated by the bearing experiment and collected vibration data including different types of faults. The results show that the accuracy rate is more superior than other six different neural network, and it is suitable for roller bearing fault diagnosis.

Forest Fire Smoke Detection Based on Multiple Color Spaces Deep Feature Fusion

The drastic increase of forest fire occurrence, which in recent years has posed severe threat and damage worldwide to the natural environment and human society, necessitates smoke detection of the early forest fire. First, a semantic segmentation method based on multiple color spaces feature fusion is put forward for forest fire smoke detection. Considering that smoke images in different color spaces may contain varied and distinctive smoke features which are beneficial for improving the detection ability of a model, the proposed model integrates the function of multi-scale and multi-type self-adaptive weighted feature fusion with attention augmentation to extract the enriched and complementary fused features of smoke, utilizing smoke images from multi-color spaces as inputs. Second, the model is trained and evaluated on part of the FIgLib dataset containing high-quality smoke images from watchtowers in the forests, incorporating various smoke types and complex background conditions, with a satisfactory smoke segmentation result for forest fire detection. Finally, the optimal color space combination and the fusion strategy for the model is determined through elaborate and extensive experiments with a superior segmentation result of 86.14 IoU of smoke obtained.

Radar antenna scanning type identification based on the fusion of multiple temporal features

Identifying the antenna scanning type (AST) of radar signal aims to deeply analyze the parameters of radar waveform and accurately judge the AST of the radar signal. Traditional methods rely heavily on domain priors and expert experience to set classification thresholds, which introduces human errors. Moreover, the generalization and anti-noise interference abilities are relatively weak, and it is difficult to deal with complex battlefield environments. The model based on the convolutional neural network (CNN) is limited by the receptive field of the convolution kernel, which makes the model unable to make full use of the global information of the signal. To fully exploit the global timing characteristics of radar signals, we propose a radar AST recognition method based on the fusion of multiple temporal features. Specifically, for the time series characteristics of the radar antenna scanning signal, a 1D CNN branch is first built to extract its short time series features. Given the limited receptive field of the convolutional network and the inability to fully consider the overall characteristics of the radar signal, we propose a transformer network branch to extract the global timing features fully. The recognition of radar AST based on multi-time series feature fusion can fully model the local and global sequence attributes, thereby improving the cognitive recognition ability of radar signals. Our method attains state-of-the-art results based on experimental findings.

Vision SLAM algorithm for wheeled robots integrating multiple sensors.

Wheeled robots play a crucial role in driving the autonomy and intelligence of robotics. However, they often encounter challenges such as tracking loss and poor real-time performance in low-texture environments. In response to these issues, this research proposes a real-time localization and mapping algorithm based on the fusion of multiple features, utilizing point, line, surface, and matrix decomposition characteristics. Building upon this foundation, the algorithm integrates multiple sensors to design a vision-based real-time localization and mapping algorithm for wheeled robots. The study concludes with experimental validation on a two-wheeled robot platform. The results indicated that the multi-feature fusion algorithm achieved the highest average accuracy in both conventional indoor datasets (84.57%) and sparse-feature indoor datasets (82.37%). In indoor scenarios, the vision-based algorithm integrating multiple sensors achieved an average accuracy of 85.4% with a processing time of 64.4 ms. In outdoor scenarios, the proposed algorithm exhibited a 14.51% accuracy improvement over a vision-based algorithm without closed-loop detection. In summary, the proposed method demonstrated outstanding accuracy and real-time performance, exhibiting favorable application effects across various practical scenarios.

AGF-Net: adaptive global feature fusion network for road extraction from remote-sensing images

Road extraction from remote-sensing images is of great significance for vehicle navigation and emergency insurance. However, the road information extracted in the remote-sensing image is discontinuous because the road in the image is often obscured by the shadows of trees or buildings. Moreover, due to the scale difference of roads in remote-sensing images, it remains a computational challenge to extract small-size roads from remote-sensing images. To address those problems, we propose a road extraction method based on adaptive global feature fusion (AGF-Net). First, a dilated convolution strip attention (DCSA) module is designed from the encoder–decoder structure. It consists of the dilated convolution and the strip attention module, which adaptively emphasizes relevant features in vertical and horizontal directions. Then, multiple global feature fusion modules (GFFM) in the skip connection are designed to supplement the decoder with road detail features, and we design a multi-scale strip convolution module (MSCM) to implement the GFFM module to obtain multi-scale road information. We compare AGF-Net to state-of-the-art methods and report their performance using standard evaluation metrics, including Intersection over Union (IoU), F1-score, precision, and recall. Our proposed AGF-Net achieves higher accuracy compared to other existing methods on the Massachusetts Road Dataset, DeepGlobe Road Dataset, CHN6-CUG Road Dataset, and BJRoad Dataset. The IoU obtained on these datasets are 0.679, 0.673, 0.567, and 0.637, respectively.