Three-dimensional Feature Research Articles

Six-Degree-of-Freedom Pose Estimation Method for Multi-Source Feature Points Based on Fully Convolutional Neural Network

An object’s six-degree-of-freedom (6DoF) pose information has great importance in various fields. Existing methods of pose estimation usually detect two-dimensional (2D)-three-dimensional (3D) feature point pairs, and directly estimates the pose information through Perspective-n-Point (PnP) algorithms. However, this approach ignores the spatial association between pixels, making it difficult to obtain high-precision results. In order to apply pose estimation based on deep learning methods to real-world scenarios, we hope to design a method that is robust enough in more complex scenarios. Therefore, we introduce a method for 3D object pose estimation from color images based on farthest point sampling (FPS) and object 3D bounding box. This method detects the 2D projection of 3D feature points through a convolutional neural network, matches it with the 3D model of the object, and then uses the PnP algorithm to restore the feature point pair to the object pose. Due to the global nature of the bounding box, this approach can be considered effective even in partially occluded or complex environments. In addition, we propose a heatmap suppression method based on weighted coordinates to further improve the prediction accuracy of feature points and the accuracy of the PnP algorithm in solving the pose position. Compared with other algorithms, this method has higher accuracy and better robustness. Our method yielded 93.8% of the ADD(-s) metrics on the Linemod dataset and 47.7% of the ADD(-s) metrics on the Occlusion Linemod dataset. These results show that our method is more effective than existing methods in pose estimation of large objects.

A method for accurate extraction of three-dimensional point cloud feature data in loading test of container ships

The accuracy of container slot dimensions directly affects the efficiency and safety of container loading and unloading. However, traditional methods of container loading tests need to be improved for their long cycles, high costs, and low production efficiency. With the application of three-dimensional laser scanning devices in ship manufacturing, the use of simulated container loading test methods based on point cloud becomes a breakthrough in key technology for container ship construction. This paper presents a multi-objective segmentation method of cargo hold feature point cloud for container ship simulated loading tests. Data preprocessing algorithms are applied to remove noise and pseudo-image, and then a Random Sample Consensus algorithm is used to remove bulkhead and cargo bottom point clouds. Based on the density-based spatial clustering of applications with noise algorithm and the region growing algorithm, an algorithm for multiple segmentation of guide rail and bottom cone point clouds is devised, and a four-plane segmentation of guide rail point clouds is realized. Finally, a method for three-dimensional point cloud data flatness assessment and a Hausdorff distance based method for calculating the guide rail lateral and longitudinal spacing are proposed. These methods are validated through the loading test of a 16,000 TEU container ship.

EEG-Based Feature Classification Combining 3D-Convolutional Neural Networks with Generative Adversarial Networks for Motor Imagery.

The adoption of convolutional neural networks (CNNs) for decoding electroencephalogram (EEG)-based motor imagery (MI) in brain-computer interfaces has significantly increased recently. The effective extraction of motor imagery features is vital due to the variability among individuals and temporal states. This study introduces a novel network architecture, 3D-convolutional neural network-generative adversarial network (3D-CNN-GAN), for decoding both within-session and cross-session motor imagery. Initially, EEG signals were extracted over various time intervals using a sliding window technique, capturing temporal, frequency, and phase features to construct a temporal-frequency-phase feature (TFPF) three-dimensional feature map. Generative adversarial networks (GANs) were then employed to synthesize artificial data, which, when combined with the original datasets, expanded the data capacity and enhanced functional connectivity. Moreover, GANs proved capable of learning and amplifying the brain connectivity patterns present in the existing data, generating more distinctive brain network features. A compact, two-layer 3D-CNN model was subsequently developed to efficiently decode these TFPF features. Taking into account session and individual differences in EEG data, tests were conducted on both the public GigaDB dataset and the SHU laboratory dataset. On the GigaDB dataset, our 3D-CNN and 3D-CNN-GAN models achieved two-class within-session motor imagery accuracies of 76.49% and 77.03%, respectively, demonstrating the algorithm's effectiveness and the improvement provided by data augmentation. Furthermore, on the SHU dataset, the 3D-CNN and 3D-CNN-GAN models yielded two-class within-session motor imagery accuracies of 67.64% and 71.63%, and cross-session motor imagery accuracies of 58.06% and 63.04%, respectively. The 3D-CNN-GAN algorithm significantly enhances the generalizability of EEG-based motor imagery brain-computer interfaces (BCIs). Additionally, this research offers valuable insights into the potential applications of motor imagery BCIs.

A three-dimensional feature-based fusion strategy for infrared and visible image fusion

Due to the lacking of attention to the scene’s essential characteristics, the existing fusion methods suffer from the deficiency of scene distortion. In addition, the lack of groundtruth can cause an inadequate representation of vital information. To this end, we propose a novel infrared and visible image fusion network based on three-dimensional feature fusion strategy (D3Fuse). In our method, we consider the scene semantic information in the source images and extract the commonality contents of the two images as the third-dimensional feature to extend the feature space for fusion tasks. Specifically, a commonality feature extraction module (CFEM) is designed to extract the scene commonality features. Subsequently, the scene commonality features are utilized together with modality features to construct the fusion image. Moreover, to ensure the independence and diversity of distinct features, we employ a contrastive learning strategy with multiscale PCA coding, which stretches the feature distance in an unsupervised manner, prompting the encoder to extract more discriminative information without incurring additional parameters and computational costs. Furthermore, a contrastive enhancement strategy is utilized to ensure adequate representation of modality information. The results of the qualitative and quantitative evaluations on the three datasets show that the proposed method has better visual performance and higher objective metrics with lower computational cost. The object detection experiments show that our results have superior performance on high-level semantic tasks.

Fault probability identification method for distribution networks based on mov-MF distribution

To address the fault identification challenge in distribution networks, a method leveraging a mixture of the von Mises–Fisher (mov-MF) distribution model for fault probability identification is proposed. Initially, the synchronous phasor measuring unit is employed to gather the post-fault steady-state voltage phase quantities, and then, the voltage phase angle values are combined to form a three-dimensional feature quantity. Subsequently, the mov-MF distribution model is initialized through the spherical K-means algorithm and the minimum message length algorithm. This model is further refined via the expectation–maximization algorithm to iteratively optimize distribution parameters. The test set data are input into the mov-MF distribution model, which has been constructed using typical fault data, to discern fault types. Finally, the efficacy of the proposed method is validated through simulation verification conducted on the IEEE 33-node distribution system. The analysis of the examples demonstrates the accuracy of the mov-MF distribution model-based fault identification method in identifying single-phase ground, two-phase ground, two-phase interphase, and three-phase short-circuit faults.

A Sheep Identification Method Based on Three-Dimensional Sheep Face Reconstruction and Feature Point Matching.

As the sheep industry rapidly moves towards modernization, digitization, and intelligence, there is a need to build breeding farms integrated with big data. By collecting individual information on sheep, precision breeding can be conducted to improve breeding efficiency, reduce costs, and promote healthy breeding practices. In this context, the accurate identification of individual sheep is essential for establishing digitized sheep farms and precision animal husbandry. Currently, scholars utilize deep learning technology to construct recognition models, learning the biological features of sheep faces to achieve accurate identification. However, existing research methods are limited to pattern recognition at the image level, leading to a lack of diversity in recognition methods. Therefore, this study focuses on the small-tailed Han sheep and develops a sheep face recognition method based on three-dimensional reconstruction technology and feature point matching, aiming to enrich the theoretical research of sheep face recognition technology. The specific recognition approach is as follows: full-angle sheep face images of experimental sheep are collected, and corresponding three-dimensional sheep face models are generated using three-dimensional reconstruction technology, further obtaining three-dimensional sheep face images from three different perspectives. Additionally, this study developed a sheep face orientation recognition algorithm called the sheep face orientation recognition algorithm (SFORA). The SFORA incorporates the ECA mechanism to further enhance recognition performance. Ultimately, the SFORA has a model size of only 5.3 MB, with accuracy and F1 score reaching 99.6% and 99.5%, respectively. During the recognition task, the SFORA is first used for sheep face orientation recognition, followed by matching the recognition image with the corresponding three-dimensional sheep face image based on the established SuperGlue feature-matching algorithm, ultimately outputting the recognition result. Experimental results indicate that when the confidence threshold is set to 0.4, SuperGlue achieves the best matching performance, with matching accuracies for the front, left, and right faces reaching 96.0%, 94.2%, and 96.3%, respectively. This study enriches the theoretical research on sheep face recognition technology and provides technical support.

Three-Dimensional Feature Tracking Study of Healthy Chinese Ventricle by Cardiac Magnetic Resonance.

Myocardial strain, as a crucial quantitative indicator of myocardial deformation, can detect the changes of cardiac function earlier than parameters such as ejection fraction (EF). It has reported that cardiac magnetic resonance(CMR) and post-processing software possess the ability to obtain the stability and repeatability strain values. Recently, the normal strain values range of people are debatable, especially in the Chinese population. Therefore, we aim to explore the ventricular characteristics and the myocardial strain values of the Chinese people by using the cardiac magnetic resonance feature tracking (CMR-FT). Additionally, we attempted to use the myocardial and chordae tendineae contours to calculate the ventricular volumes by the CMR-FT. This study may provide valuable insights into the application of CMR-FT in tracking the ventricular characteristics and myocardial strain for Chinese population, especially in suggesting an referable myocardial strain parameters of the Chinese. A total of 109 healthy Chinese individuals (age range: 18 to 58 years; 52 males and 57 females) underwent 3.0T CMR to acquire the cardiac images. The commercial post-processing software was employed to analyse the image sequence by semi-automatic processing, then the biventricular morphology (End-Diastolic Volume, EDV; EDV/Body Surface Area, EDV/BSA), function(EF; Cardiac Output, CO; Cardiac Index, CI) and strain(Radial Strain, RS; Circumferential Strain, CS; Longitudinal Strain, LS) values were obtained.The biventricular myocardial strain values were stratified according to the age and gender. The Left Ventricular( LV base, mid, apex) and myocardial strain values of three coronary artery areas were calculated based on the the strain value of LV American Heart Association(AHA) 16 segments. It was shown that the females had larger LV globe strain values compared with the males (LVGPRS: 42.0 ± 8.5 versus 33.6 ± 6.2%, P < 0.001; LVGPCS: -21.2 ± 2.1 versus - 19.7 ± 2.3%, P < 0.001; LVGPLS: -16.4 ± 2.6 versus - 14.6 ± 2.2%, P < 0.001;). Moreover, the differences in RS, CS, and LS among the LV myocardium 16 segments were obvious. However, the right ventricle (RV) strain values showed non-normal distribution in the volunteers of this research. Here, we successfully tracked the characteristics of bilateral ventricles in healthy Chinese populations through using the 3.0T CMR. We confirmed that there was a gender difference in LV Globe Strain values. In addition, we obtained strain values for each myocardial segment of the LV and different coronary artery regions based on the AHA 16 segments method, Our results also showed that the RV strain values with a non-normal distribution, and RV global strain values were not related to the gender and age. Furthermore, LVGPRS, LVGPLS, and RVGPRS were significantly correlated with BMI, CO, CI, and EDV in the Chinese population.

How many specimens make a sufficient training set for automated three-dimensional feature extraction?

Deep learning has emerged as a robust tool for automating feature extraction from three-dimensional images, offering an efficient alternative to labour-intensive and potentially biased manual image segmentation methods. However, there has been limited exploration into the optimal training set sizes, including assessing whether artficial expansion by data augmentation can achieve consistent results in less time and how consistent these benefits are across different types of traits. In this study, we manually segmented 50 planktonic foraminifera specimens from the genus Menardella to determine the minimum number of training images required to produce accurate volumetric and shape data from internal and external structures. The results reveal unsurprisingly that deep learning models improve with a larger number of training images with eight specimens being required to achieve 95% accuracy. Furthermore, data augmentation can enhance network accuracy by up to 8.0%. Notably, predicting both volumetric and shape measurements for the internal structure poses a greater challenge compared with the external structure, owing to low contrast differences between different materials and increased geometric complexity. These results provide novel insight into optimal training set sizes for precise image segmentation of diverse traits and highlight the potential of data augmentation for enhancing multivariate feature extraction from three-dimensional images.

Three-dimensional feature tracking cardiac magnetic resonance in hypertrophic cardiomyopathy: comparison with two-dimensional algorithm

Three-dimensional feature tracking cardiac magnetic resonance in hypertrophic cardiomyopathy: comparison with two-dimensional algorithm

DiaTracer enables spectrum-centric analysis of diaPASEF proteomics data.

Data-independent acquisition (DIA) has become a widely used strategy for peptide and protein quantification in mass spectrometry-based proteomics studies. The integration of ion mobility separation into DIA analysis, such as the diaPASEF technology available on Bruker's timsTOF platform, further improves the quantification accuracy and protein depth achievable using DIA. We introduce diaTracer, a new spectrum-centric computational tool optimized for diaPASEF data. diaTracer performs three-dimensional (m/z, retention time, ion mobility) peak tracing and feature detection to generate precursor-resolved "pseudo-MS/MS" spectra, facilitating direct ("spectral-library free") peptide identification and quantification from diaPASEF data. diaTracer is available as a stand-alone tool and is fully integrated into the widely used FragPipe computational platform. We demonstrate the performance of diaTracer and FragPipe using diaPASEF data from cerebrospinal fluid (CSF) and plasma samples, data from phosphoproteomics and HLA immunopeptidomics experiments, and low-input data from a spatial proteomics study. We also show that diaTracer enables unrestricted identification of post-translational modifications from diaPASEF data using open/mass offset searches.

Synthesis of Azo-Substituted Bicyclo[1.1.1]pentanes (BCPs) via Base-Promoted Halogen Atom Transfer.

Because of the three-dimensional bioisosteric feature, bicyclo[1.1.1]pentylamines (BCPAs) are valuable scaffolds in synthetic chemistry and medicinal chemistry. Here, we report a Halogen Atom Transfer (XAT) mediated radical C-N coupling between C3-iodo-BCPs and diazonium salts in the presence of base. Similarly, a multicomponent reaction (MCR) enables the simultaneous construction of the C-C bond and C-N bond simultaneously. Versatile roles of diazonium salts were also explored.

Multi-scale 3D-CRU for EEG emotion recognition* *This research has been supported by the National Natural Science Foundation of China Joint Fund Key Project (U2003207), National Natural Science Foundation of China (6220100261902064), Natural Science Foundation of Anhui Province (No. 1908085MF209), Key Projects of Natural Science Foundation of Anhui Province Universities (No. KJ2018A0018).

In this paper, we propose a novel multi-scale 3D-CRU model, with the goal of extracting more discriminative emotion feature from EEG signals. By concurrently exploiting the relative electrode locations and different frequency subbands of EEG signals, a three-dimensional feature representation is reconstructed wherein the Delta (δ) frequency pattern is included. We employ a multi-scale approach, termed 3D-CRU, to concurrently extract frequency and spatial features at varying levels of granularity within each time segment. In the proposed 3D-CRU, we introduce a multi-scale 3D Convolutional Neural Network (3D-CNN) to effectively capture discriminative information embedded within the 3D feature representation. To model the temporal dynamics across consecutive time segments, we incorporate a Gated Recurrent Unit (GRU) module to extract temporal representations from the time series of combined frequency-spatial features. Ultimately, the 3D-CRU model yields a global feature representation, encompassing comprehensive information across time, frequency, and spatial domains. Numerous experimental assessments conducted on publicly available DEAP and SEED databases provide empirical evidence supporting the enhanced performance of our proposed model in the domain of emotion recognition. These findings underscore the efficacy of the features extracted by the proposed multi-scale 3D-GRU model, particularly with the incorporation of the Delta (δ) frequency pattern. Specifically, on the DEAP dataset, the accuracy of Valence and Arousal are 93.12% and 94.31%, respectively, while on the SEED dataset, the accuracy is 92.25%.

Wind Shear Operation-Based Competency Assessment Model for Civil Aviation Pilots

Wind Shear Operation-Based Competency Assessment Model for Civil Aviation Pilots

Single crystalline molybdenum nanostructure for generating reflective micro-focal spot X-rays: Structure-controllable preparation and characteristic-selectable radiation

The use of properly designed nanostructured anode targets enables direct generation of high performance X-rays. In particular, this scheme has attracted widespread research interest for obtaining micro-focal spot X-ray radiation. In this study, we developed a thermal evaporating deposition technique and achieved controllable preparation of different single crystalline molybdenum (Mo) nanostructures by exploiting the lattice matching relationship between the substrate and the product crystal orientation. When used as an anode target for a reflective X-ray radiation device, these nanostructured Mo single crystals exhibit advantages over traditional targets at larger radiation angles. Further experimental and simulation results demonstrate the geometric characteristics of these Mo nanostructures and their correlation with the radiation intensity, focal spot size, and brightness of the generated X-rays. Among them, a unique frustum of square pyramid single crystalline Mo nanostructure stands out, due to its combination of three-dimensional body feature of nanostructures and radiation plane feature of planar targets. The research results not only provide a class of nanostructured anode target materials for reflective X-ray sources with different radiation angle requirements, but also make it possible to design and develop miniaturized devices for X-ray sources with selectable radiation performance and micro-focal spot.

Characterization of road surface topography changes during polishing process with characteristic roughness parameter

In this paper, three-dimensional structure feature points of the asphalt mixture were obtained based on the industrial CT. The surface morphology of the asphalt mixture was then extracted through image morphology processing and evolutionary matching algorithms. A series of accelerated polishing tests were utilized to simulate the long-term polishing effect of the tire load on the asphalt pavement. Combining with digital image analysis, the full-life morphology evolution trend of the asphalt pavement was analyzed, and the full-life surface morphology datasets of the asphalt pavement were constructed. By comparing the power spectrum density of the asphalt pavement morphology at different wear stages, a segmented evaluation method based on the spectral fractal method was proposed. Combining the spectral fractal parameters and statistical geometric parameters, a multi-scale asphalt pavement surface morphology feature evaluation index was proposed. This provides a more intricate and detailed analysis of the surface structure of asphalt mixtures.

Cavitation fault diagnosis of centrifugal pump based on RIME-SDAE

In order to fully extract the fault characteristics of different cavitation fault states of centrifugal pumps, reduce the influence of hyperparameter settings on the identification and classification results of machine learning algorithms, this paper designs a network model based on the rime optimization algorithm (RIME) to improve the stacked denoising autoencoder (SDAE) network, which is named as RIME-SDAE. First of all, Singular Value Decomposition (SVD) is used to denoise the X, Y and Z signals of the triaxial vibration sensor, and time-domain, frequency-domain and time-frequency features are extracted to construct a signal feature set. Secondly, the three-dimensional feature indicators are analyzed and selected to be merged into the input dataset, and SDAE is trained with the RIME is used to determine the model parameters of SDAE synchronously. The feasibility of the proposed method is verified by the signals collected in the actual test, and the test results show that the accuracy on the test set reaches more than 98 %.

Reverse calculation and quantitative estimation of JRC3D for rough rock fracture surfaces in various shear directions

The quantitative evaluation of the three-dimensional roughness feature of natural rock fracture surface is essential for predicting shear strength. This article presents an effective approach to estimate the three-dimensional roughness of natural fracture surfaces from the two-dimensional roughness of fracture profile. Firstly, four types of rough fracture samples were created using rock-like materials, followed by conducting direct shear tests of the fracture surfaces under 4 different normal pressures and 4 shear directions. The results of the shear strength tests exhibited a distinct discrete pattern and consistently increased with higher normal pressures. Furthermore, the distribution of the damage area on the fracture surface after shear was significantly impacted by the distribution and size of asperities, as well as the shear direction. The three-dimensional roughness of four fracture surfaces in different directions was first reverse calculated using the shear strength model. Using the weighted positive angle, the estimation of the three-dimensional roughness in different directions of the fracture surface was then determined from the two-dimensional roughness of fracture profiles using both the arithmetic mean method and the weighted average method. The weighted average method effectively captures the contribution of the two-dimensional roughness of fracture profiles to the three-dimensional roughness of fracture surfaces, and exhibits a lower estimation error compared to the arithmetic mean method. Compared to other roughness indicators, the three-dimensional roughness evaluated through the weighted positive angle indicator are closest to the experimental result. Finally, the influence of transverse spacing on the estimated three-dimensional roughness was discussed.

EEG Emotion Recognition Based on Self-Distillation Convolutional Graph Attention Network

A convolution graph attention model based on self-distillation convolutional graph attention network (SDC-GAT) is proposed for multi-channel EEG emotion recognition. Firstly, two-dimensional feature matrix based on EEG time-domain features are constructed, and the matrix is fed into the graph attention neural network to learn the internal connections between electrical brain channels located in different brain regions. Meanwhile, the three-dimensional feature matrix is constructed according to the relative positions of the electrode channels, and the self-distillation network is employed to extract local high-level abstract features containing electrode spatial position information from the three-dimensional feature matrix. Finally, outputs of the two networks are integrated to determine the emotional states. Experiments were performed on the DEAP dataset. The experimental results show that the spatial domain information of the electrode channel and the internal connection relationship between different channels are beneficial for emotion recognition. In addition, the proposed model can effectively fuse these information to improve the performance of multi-channel EEG emotion recognition.

UAV hyperspectral analysis of secondary salinization in arid oasis cotton fields: effects of FOD feature selection and SOA-RF.

Secondary salinization is a crucial constraint on agricultural progress in arid regions. The specific mulching irrigation technique not only exacerbates secondary salinization but also complicates field-scale soil salinity monitoring. UAV hyperspectral remote sensing offers a monitoring method that is high-precision, high-efficiency, and short-cycle. In this study, UAV hyperspectral images were used to derive one-dimensional, textural, and three-dimensional feature variables using Competitive adaptive reweighted sampling (CARS), Gray-Level Co-occurrence Matrix (GLCM), Boruta Feature Selection (Boruta), and Brightness-Color-Index (BCI) with Fractional-order differentiation (FOD) processing. Additionally, three modeling strategies were developed (Strategy 1 involves constructing the model solely with the 20 single-band variable inputs screened by the CARS algorithm. In Strategy 2, 25 texture features augment Strategy 1, resulting in 45 feature variables for model construction. Strategy 3, building upon Strategy 2, incorporates six triple-band indices, totaling 51 variables used in the model's construction) and integrated with the Seagull Optimization Algorithm for Random Forest (SOA-RF) models to predict soil electrical conductivity (EC) and delineate spatial distribution. The results demonstrated that fractional order differentiation highlights spectral features in noisy spectra, and different orders of differentiation reveal different hidden information. The correlation between soil EC and spectra varies with the order. 1.9th order differentiation is proved to be the best order for constructing one-dimensional indices; although the addition of texture features slightly improves the accuracy of the model, the integration of the three-waveband indices significantly improves the accuracy of the estimation, with an R2 of 0.9476. In contrast to the conventional RF model, the SOA-RF algorithm optimizes its parameters thereby significantly improving the accuracy and model stability. The optimal soil salinity prediction model proposed in this study can accurately, non-invasively and rapidly identify excessive salt accumulation in drip irrigation under membrane. It is of great significance to improve the growing conditions of cotton, increase the cotton yield, and promote the sustainable development of Xinjiang's agricultural economy, and also provides a reference for the prevention and control of regional soil salinization.

Effect of wall curvature on heat transfer and hydrodynamics in a ribbed cooling passage

Simplified rectangular ribbed cooling passages with a flat wall are extensively considered in exploring the internal cooling features of turbine blades, but the realistic blade has a twisted shape inherently. The effects induced by the curved wall have not been clarified in detail. In this work, adopting a verified v2f turbulence model, numerical investigations are completed to evaluate the effects of the curved wall on the internal cooling characteristics of a ribbed channel. Adopting the unified ribbed channel, flat, convex, and concave walls with distinct curvatures are comprehensively evaluated and compared in a wide Re range for the turbulent flow and heat transfer features as well as the flow and thermal performance. It is found that using the flat wall, ribs can typically induce recirculation vortices having a two-dimensional nature. In contrast, the curved wall significantly contributes to the counter-rotating vortex pairs on the spanwise plane. Combined with recirculation vortices offered by the ribs, the turbulent flow of the cooling channel with the curved wall has a remarkable three-dimensional feature. Hence, the turbulent activity and fluid mixing are enhanced greatly along with the raised heat transfer enhancement and friction loss. Particularly, the convex wall with a curvature of K = 4 provides 28.6 % higher heat transfer performance (Nu/Nu0) but 88.4 % higher resistance (f/f0) than the flat wall. Considering the overall cooling performance, the concave wall with a relatively small curvature is suggested with an improvement of up to 32.8 % concerning the factor (Nu/Nu0)/(f/f0) and 9.5 % on (Nu/Nu0)/(f/f0)1/3. Finally, it is highlighted that considering the effect of the wall curvature, the current study stimulates the mechanistic understanding and provides a design guideline for high-performance blade internal cooling.