Search

One of the key challenge faced by space-based network is how to maximize the demand for on-board resources for ground communication tasks, given the limited availability of satellite resources. For this challenge, firstly, we propose a joint state space of satellite task requirements and resource pools to obtain the global information of the environment, avoiding convergence to local optimal strategies. Secondly, we propose a new joint partitioning method for frequency and time resources, which avoids the fragmentation of the resource to the maximum extent. Thirdly, a new algorithm called dynamic weight based soft actor critic (DWSAC) is proposed, which enhances the update range when the actions taken by the agent significantly contribute to the improvement of system performance, otherwise weakens the update range, significantly improving the convergence efficiency and performance of the soft actor critic (SAC). The results show that the proposed model and algorithm have good practicability, which can make the average resource occupancy rate higher and the running cost lower.

There are a large category of textile images that have the characteristics of high local feature repetition rate and complex background information. These types of images have significant intra-class differences and small inter-class differences, making it impossible to perform classification training. Making it difficult for existing methods to accurately retrieve textile images. To improve the retrieval accuracy of textile images, this paper defines multiple repeated local fine-grained features in textile images as textile image “feature components”, extracts multiple “feature components” from a textile image, and fuses the “feature components” to generate the definition of textile “fingerprints”. We propose an image retrieval method based on a pre-trained Mask R-CNN model to extract multiple “feature components” in the textile image, then extract the depth feature of the textile again through a convolution neural network, and fuse the extracted depth feature to obtain the “fingerprint” of the textile. The obtained “fingerprint” can effectively eliminate the interference of large area background and a large number of local repeated features in the textile image, and improve the efficiency of textile retrieval. A series of comparative experiments are carried out on textile image data sets with repeated features. The experimental results show that the proposed method is generally effective.

Cross-scene knowledge transfer has been proven effective to deal with the small-sample problem in hyperspectral image (HSI) classification. Currently, most of the existing works are based on the homogeneous settings where the source and target HSIs are observed from the same feature space, while the heterogeneous situation, which is more common in real-world applications, is under insufficient exploration. In this article, we propose a novel transfer learning approach for HSI classification, which is capable of transferring discriminative spectral-spatial feature between heterogeneous datasets. Specifically, we first extract spectral-spatial features of source and target scenes by applying the multiscale convolutional sparse decomposition (MCSD) method. By performing MCSD, the spectral information and spatial structure information at different scales can be jointly adapted to learn transferable features for classification. Then, in order to overcome the heterogeneity between the two feature sets, we build a structure preserved distribution alignment (SPDA) model to learn domain-specific projections to map the feature samples into a shared latent subspace where the discriminative knowledge can be effectively transferred. With proper reformulation, we give the analytical solution of the objective function and generate an optimization approach to solve the SPDA model efficiently. Experiments conducted on several real data pairs demonstrate that the proposed approach can explicitly narrow the disparity between heterogeneous HSIs, and yield superior classification results compared with other representative heterogeneous transfer learning methods.

The distinction of similar classes has always been the core issue in image classification. In this article, a new hierarchical classification process based on three-dimensional attention soft augmentation (HC-3DAA) is proposed to improve the accuracy of classifiers, especially for the accuracy between similar classes. In HC-3DAA processing, the merging matrix is first constructed based on the validation confusion matrix to measure the similarity among different classes. Specifically, the 3-D attention soft augmentation module combined with CutMix is designed for guiding the network model to focus on the key discriminative features. Then, the extracted 3-D feature differences between similar classes are inserted into the attention module for the reclassification to get higher classification accuracy. To evaluate the performance of HC-3DAA, CutMix models with different feature dimensions and the HC module are separately discussed on two widely used hyperspectral datasets. Two different classifiers 3-D convolutional neural network and ResNet are included in the comparative analysis. Besides, experimental results also demonstrate that the proposed HC-3DAA outperforms several state-of-the-art attention-based methods.

The compressive sensing (CS)-based optical remote sensing (ORS) imaging system has been verified for the feasibility through numerical simulation experiments. The CS-based ORS imaging system can reduce the demand for sampling equipment, reduce sampling data, save storage space, and reduce transmission costs. However, it needs to reconstruct the original scene when facing the task of ship detection. The scene reconstruction process of CS is computationally expensive, high memory demanding, and time-consuming. In response to this problem, this article proposes an innovation pipeline to perform ship detection tasks, i.e., directly performing ship detection on CS measurements obtained by the imaging system, which avoids the process of scene reconstruction. To achieve the ship detection of CS measurements in the pipeline, we design a convolutional neural network-based algorithm, CS-CenterNet, which jointly optimizes the scene compression sampling phase and the measurements’ ship detection phase. CS-CenterNet is divided into convolution measurement layer (CML), optimized hourglass network (OHgN), and optimized three-branch head network (OTBHN). First, CML without bias or activation function simulates the block compression sampling process in CS-based ORS imaging system, which performs convolutional coding on the scene to obtain the measurements. Second, OHgN extracts high-resolution feature information of measurements. Finally, OTBHN performs heat-map prediction, center-point offset prediction, and width–height prediction. We test the performance of CS-CenterNet using the HRSC2016 and LEVIR datasets. The experimental results show that the algorithm can achieve high-accuracy ship detection based on CS measurements of ORS scenes.

With the development of society and the advancement of science and technology, artificial intelligence has also emerged as the times require. In computer vision, deep learning based on convolutional neural networks(CNN) achieves state-of-the-art performance. However, the massive data requirements of deep learning have long been a pain point in the field, especially in the medical field, where it is often difficult (and sometimes impossible) to obtain enough training data for some specific tasks. To overcome insufficient and unbalanced data, in this paper, we focus on the generation and balance of data on radiation-induced pneumonia, an extremely rare disease with a low incidence. As a result, datasets on this disease are extremely sparse and unevenly distributed. To address the above problems, the predecessors’ method is often to use generative models to generate data as a complement of the fewer samples to achieve a balanced distribution of data samples. Among various generative models, CycleGAN is widely used in medical image generation due to its cycle consistency to achieve style migration without changing the basic content. However, the original CycleGAN method has many shortcomings, especially in Few-shot and the data unevenly distributed, its performance will be greatly reduced. To make the generated data samples retain the original structure and have finer and clearer details, this paper proposes a mask-based self-attention CycleGAN data augmentation method. A self-attention branch is added to the generator and two different loss functions named Self-Attention Loss and Mask Loss are designed. To stabilize the training process, spectral normalization is introduced to improve the discriminator and MS-SSIM and L1 joint loss are used to improve the original identity loss. The ResNet18 is used to complete classification experiments on the radiation-induced pneumonia dataset and the COVID-19 dataset respectively. Four classification performance indicators: the area under the ROC curve (AUC), Accuracy (ACC), Sensitivity (SEN), and Specificity (SPE) are calculated to verify the effectiveness and generalization of our method. Compared with the original CycleGAN and traditional data augmentation, the classifier trained by data augmentation using our method has outstanding performance in multiple classification indicators and has better classification performance. Experimental results show that our method solves the problem of insufficient samples and data imbalance in the pneumonia dataset by generating high-quality pneumonia images. Code is available at https://github.com/ngfufdrdh/CycleGAN-lung.

The physical optics (PO) method treats the outer surface of airborne radomes in front of antennas as a secondary source region when analyzing the electromagnetic performance. The effective reduction of the secondary source region can significantly improve the computational efficiency, which is helpful for the rapid analysis and design of airborne radomes. In this paper, we present a cylindrical equivalent source-based PO method by introducing an antenna radiation cylinder that can cover the main near-field radiation characteristics of the antenna. The cubic-spline interpolation technique is used to standardize the solution method for the boundary of the cylindrical equivalent source of different antenna-radome systems. The results of a tangent-ogival radome verify the validity of the proposed method. Compared with the conical equivalent source-based PO method, the proposed method improves the efficiency by 71.83%. It can be applied to airborne antenna-radome systems with antenna diameters of 12.14 times of wavelength and above, using 10% as the error threshold.

Hyperspectral anomaly detection has been a hot topic in the field of remote sensing due to its potential application prospects. However, anomaly detection still has two typical problems to be solved. First, the target in the hyperspectral image is usually mixed with the background. Thus, the background is often misjudged as a target without an effective background suppression strategy resulting in false alarms. Second, the non-Gaussian distribution of the complex background affects the judgment of the detector on the abnormality of the target and leads to missed detection. This article proposes an anomaly detection method based on the optimal kernel and high-order moment correlation representation. In selecting the kernel function, the advantage of the Gaussian kernel in anomaly detection is demonstrated through the analysis of the signal uncertainty principle. On this basis, the optimal kernel is determined with sufficient background suppression by isolation forest. In addition, to reduce the influence of non-Gaussian distribution data on detection, the proposed method adopts high-order moments for statistical correlation representation, which further enhances the separability of the background and target after kernel mapping. The experimental results show that the proposed method can effectively suppress the background. Furthermore, it is confirmed that the Gaussian kernel function is effective in improving anomaly detection accuracy. Moreover, the high-order moment correlation representation can highlight the target after kernel mapping, thereby reducing the false alarm rate and obtaining a better detection result.

A novel network model “bidirectional depth residuals gated recurrent unit network (BDR-GRU) ” is designed and implemented for improving the effectiveness of Image Captioning. BDR-GRU is designed based on encoder and decoder architecture. Moreover, the network can run on an NVIDIA JETSON TX2 processor, which makes the algorithm applied to mobile robots. In the encoding stage, the convolution neural network is used to obtain the multi-dimensional vector information of the image, and the BDR-GRU network is used to complete the sentence generation in the decoding stage. The BDR-GRU network model is a new recurrent neural network model, which is improved on the basics of the GRU network. Firstly, the layer of the GRU network is increased from a single layer to multiple layers. Secondly, the bidirectional derivation structure is redesigned to enhance the ability of derivation. Finally, the residual mechanism between levels is designed to prevent the disappearance of gradient and over-fitting caused by the increase of the layers. Experiments are carried out on TX2 processor and have been done to verify the effectiveness of our design, and the results are compared with img-gLSTM network model, neural talk model, attention model, and unidirectional GRU model, then the results are analyzed. The experimental results show that the CIDEr evaluation value of our network model is 12.7% higher than that of the img-gLSTM network and 14.6% higher than that of the Neural Talk network, other evaluation indicators also improve significantly. The experimental results prove the significance of our BDR-GRU model.

With the increase in the availability of multitemporal hyperspectral images (HSIs), HSIs change detection (CD) methods, including pixel-level and subpixel-level based methods, have attracted great attention in recent years. However, the widespread presence of mixed pixels in HSIs may make it difficult for pixel-level methods to detect subtle changes; meanwhile, the less utilization of spatial information may also lead to limitations in some subpixel-level methods. Therefore, a joint framework, which aims to combine the advantages of pixel-level in spatial utilization and subpixel-level in temporal and spectral exploration, is proposed to enhance the performance of HSIs CD. Two models, convolutional sparse analysis and temporal spectral unmixing, are introduced and presented to characterize different spatial structures and overcome the effects of spectral variability under this framework, respectively. In addition, a multiple CD-based on subpixel analysis is discussed as well. Experiments conducted on three bitemporal HSIs datasets indicate that the proposed framework is robust in capturing effective features and has achieved great detection accuracy.

The retrieval of land surface temperature (LST) using thermal infrared (TIR) data is important in many applications. However, TIR data usually suffer from low spatial resolution. We introduce a novel subpixel LST estimation model using the information-transfer-based adaptive ensemble extreme learning machine (IT-AEELM). The proposed method constructs a reliable relationship between subpixel LST and the input high-resolution visible and near-infrared (VNIR) data, short-wave infrared (SWIR) data, and low-resolution TIR data. Based on a detailed analysis of different ground objects, we divide the input data into multiple subsets. Instead of using consistent land surface parameters (LSPs), we utilize different LSPs to characterize the land surface properties in each subset. The VNIR-SWIR-LSPs data and the low-resolution LST are used to train a novel IT-AEELM network, where a feedback ensemble learning scheme is introduced to effectively remove inaccurate estimates. The main difference of the model against existing methods is that it builds a robust architecture at different spatial scales, which provides benefits including lower demand for training data, more rapid and accurate acquisition of subpixel LST, and better adaption to heterogeneous land surface. Numerical experiments demonstrate that the proposed method significantly improves the accuracy of subpixel LST compared with the state-of-the-art algorithms.

Recently, deep learning-based methods outperform others in hyperspectral image (HSI) classification. However, the deep learning methods require sufficient labeled samples to improve performance, which is unfeasible in practice. The training labels are usually limited in HSIs that need to be classified (namely target domain), while other available labels in multisource HSIs (namely source domain) are not utilized effectively. To mitigate these issues, an attention multisource fusion method of few-shot learning (AMF-FSL) is proposed for small-sized HSI classification. AMF-FSL is an implementation of few-shot learning (FSL) in the meta-learning field, which can transfer the learned ability of classification from multiple source data to target data. The process of learning to classify in AMF-FSL is not restricted by the traditional requirement of the same distribution between the source and target domains, which can learn from the source domain and apply it to a different distribution in the target domain. Moreover, the multisource domain adaption in AMF-FSL has the capacity of extracting features from fused homogeneous and heterogeneous data in the source domain, which can improve the generalization of the classification model in the cross domains. Specifically, the multisource domain adaption contains three modules, namely the target-based class alignment, domain attention assignment, and multisource data fusion, which are responsible for aligning the class space, paying band-level attention, and merging the distributions of homogeneous and heterogeneous data in the source domain. The experimental results demonstrate the effectiveness of the multisource domain adaption and the superiority of AMF-FSL over other state-of-the-art methods in small-sized HSI classification.

In recent years, semisupervised spectral–spatial feature extraction (FE) methods for hyperspectral image (HSI) classification have shown promising performance by combining spectral–spatial information and label information. A problem that has not been addressed satisfactorily is that how to effectively and collaboratively use this abundant information contained in the hyperspectral data to exhibit better HSI classification performance. In this article, a novel FE method based on joint spectral–spatial information is proposed for HSI classification, which consists of the following steps. First, an effective re-expression for the original data is constructed by incorporating texture features extracted by extended multiattribute profiles with the original HSI. Thus, every pixel can be described by diverse and complementary information in the spectral–spatial domain. Then, the improved neighborhood preserving embedding (NPE) is proposed to establish a relatively accurate reconstruction model and mine high-reliable neighborhood structure from a global perspective by a new distance metric, which incorporates spectral bands, texture features, and geographical information simultaneously. Finally, the low-dimensional and high-discriminative features for HSI classification are obtained by combining the scatter matrices of local fisher discriminant analysis based on labeled samples and the improved NPE based on the whole data. Experimental results on three real-world HSI datasets show that the proposed method can effectively utilize both the label information and the spectral–spatial information, and hence, achieve much better classification performance compared to the conventional FE methods and some state-of-the-art spectral–spatial classification methods.

Coal water inrush acoustic emission (AE) signal is characterized by time varying, nonstationary, unpredictable and transient properties. To extract effective features representing coal water inrush information, the AE signal is analyzed by the wavelet characteristic energy spectrum coefficient based on wavelet theory. The feasibility of the wavelet feature coding has confirmed from code scheme's availability and consistency, and it proves that the coding method can be used as a sign of waveform identification. The inclusion of energy distribution characteristics makes the waveform features more ordered and simplified. While the analysis of the obtained feature encoding in chronological order, it is possible to obtain the state of the time series signals, to lay an important basis for analyzing the evolution of water inrush acoustic emission coal from the time-series level, such that a change dynamic characteristic acoustic emission signal becomes possible. And this will lay an important foundation for the time sequence analysis of acoustic emission event's evolution in mine water inrush, reduces the trans-mission of invalid signals and improves the efficiency of downhole communication.

Most image fusion algorithms based on hyperspectral unmixing perform worse with the lower spatial resolution of hyperspectral image (HSI) for the reason that the estimated endmembers and abundance deviate from the truth value. Therefore, it is more meaningful to unmix the low spatial resolution hyperspectral image (LRHSI) accurately, which is also helpful to improve the image fusion performance. In order to enhance the spatial resolution of LRHSI, this article proposes an alternating direction iterative nonnegative matrix factorization (ADINMF) based on linear hyperspectral unmixing algorithm. It takes multispectral image as a constraint to improve the spatial resolution of LRHSI. First, we use blind source separation to initialize the endmember and abundance of hyperspectral and multispectral images, respectively. Then, we alternately update the endmembers and abundance in the framework of nonnegative matrix factorization by multiplication iterative algorithm. The updated endmembers and abundance are constrained to each other. We compare the experimental results of simulated dataset and three groups of real datasets. Experimental results show that the proposed method not only accurately extracts the endmembers of LRHSI, but also obtains a significant fusion performance improvement.

To avoid using a large 4D-Hough counting space (HCS) and complex invariant features of generalized Hough transform (GHT) or its extensions when detecting objects in remote sensing image (RSI), a tensored GHT (TGHT) is proposed to extract object contour by simple gradient angle feature in a 2D-HCS using a single training sample. Considering that tensor can record the structure relationship of object contour, tensor representation R-table is constructed to record the contour information of template. For slice centered at each position of RSI, the tensor-space-based voting mechanism is presented to use the tensor that records the contour information of slice to gather votes at the same entry of 2D-HCS. Furthermore, a multiorder binary-tree-based searching method is presented to accelerate voting by searching the index numbers of elements in tensors. In addition, by solving the tensor-space-based optimization problem that is used to determine the candidates objects, the cause of false alarms (FAs) caused by interferences with complex contour and FAs caused by interferences that are partial-similar to objects is revealed, and the matching rate and matching sparsity-based strategies are then proposed to remove these FAs. Using public RSI datasets with different scenes, experimental results demonstrate that TGHT reduces nearly 99% storage requirement compared with GHT for RSI with size exceeding 1000 × 1000 under small time consumption, and outperforms the well-known contour extraction methods and state-of-the-art deep-learning-based methods in terms of precision and recall.

The small size of labeled samples has always been one of the great challenges in hyperspectral image (HSI) classification. Recently, cross-scene transfer learning has been developed to solve this problem by utilizing auxiliary samples of a relevant scene. However, the disparity between hyperspectral datasets acquired by different sensors is a tricky problem which is hard to overcome. In this article, we put forward a cross-scene deep transfer learning method with spectral feature adaptation for HSI classification, which transfers the effective contents from source scene to target scene. The proposed framework contains two parts. First, the distribution differences of spectral dimension between source domain and target domain are reduced through a joint probability distribution adaptation approach. Then, a multiscale spectral-spatial unified network with two-branch architecture and a multiscale bank is designed to extract discriminating features of HSI adequately. Finally, classification of the target image is achieved by applying a model-based deep transfer learning strategy. Experiments conducted on several real hyperspectral datasets demonstrate that the proposed approach can explicitly narrow the disparity between HSIs captured by different sensors and yield ideal classification results of the target HSI.

Due to the different spatial properties presented by various ground objects in hyperspectral image (HSI), multiscale-based feature extraction approaches have been developed for HSI classification in recent years. However, the spatial features of different scales are usually acquired at the cost of obscuring the structural information of input image, which severely limits the effectiveness of multiscale strategy. In this article, a convolutional sparse decomposition (CSD) model is introduced to characterize the significant spatial structures of hyperspectral data while removing the irrelevant noise and local textures at the specific scale. Based on the CSD model, a multiscale spectral-spatial feature extraction framework is generated, which consists of the following steps. First, the spectral dimensionality of the original HSI is reduced through a segmented averaging approach. Second, spatial features at different scales are separated from the dimension-reduced data by solving the CSD model with different regularization parameters. Finally, principal component analysis is performed and the obtained multiscale spectral-spatial features are stacked together for classification. Experiments conducted on three widely used hyperspectral datasets demonstrate that the proposed method is robust in capturing effective features of ground objects at different scales and leads to better classification results than several state-of-the-art methods.

The aluminum strands break and surface damage will gradually develop into irreversible destruction to the conductors and the whole transmission lines, and even lead to serious power outages. In order to find out the potential strands break and damage faults and prevent its further deterioration, a recognition method of conductor break and surface defects in transmission lines' unmanned aerial vehicle (UAV) inspection is presented in this paper. First, a conductor image is obtained by the UAV image acquisition system, and then, the conductor region is extracted by the adaptive threshold segmentation after the enhancement processing by the gray variance normalization method (GVN). Second, the conductor break is detected by the square wave transformation (SWT) of its grayscale distribution curves, which is simple and effective. Meanwhile, the conductor surface defects are identified by the projection algorithm of the GVN image of the conductor region. Finally, calculating the number of broken strands and filtering the suspect defects, the final fault diagnosis results can be obtained. We analyze the performance of the technology by a series of experiments, and the results demonstrate that the proposed method can measure the conductor break and surface defects faults with the average accuracy of 90.45% and 92.05%, respectively. The health condition of the conductor can be monitored based on the method presented in this paper, by which the safe operation of transmission lines can be guaranteed.

Hydrophobicity of the composite insulators is a parameter which is closely related to the insulation properties. In order to determine whether the insulator in the transmission line needs to be withdrawn, it is necessary to objectively and accurately determine the hydrophobicity grade of the composite insulator. The study on hydrophobicity detection of composite insulators of transmission lines by image analysis is proposed in this paper, by which the hydrophobicity of composite insulators can be measured timely. The grayscale, median filtering, and histogram equalization of the hydrophobic images are fulfilled sequentially, and the water drops (or water marks) region is segmented by binary large object analysis and four parameters related to water repellency are extracted. In addition, the improved support vector machine algorithm is adopted to calculate the hydrophobicity of composite insulators. The results show that seven typical hydrophobic levels of insulators can be recognized by the proposed hydrophobicity detection method whose accuracy is more than 91%.

The performance of a speaker recognition system depends highly on which acoustic features are used. Most speaker recognition systems use short-term acoustic features extracted from a single speech frame, and the most popular short-term acoustic features are the Mel-frequency cepstral coefficients (MFCCs). The short-term features are generally static features no dynamic information in the speech signal is included in either cepstral coefficients or an MFCCs frame. Using an analysis sparse representation model, in this paper, we introduce the long-term acoustic (LTA) feature for text-independent speaker recognition, which is a sparse presentation of the static features and dynamic information for the speaker's speech. First, the speech signal is segmented into frames which are overlapping with each other, and then the MFCCs frame features can be extracted to construct some super MFCCs frames by stacking some following frames of the current frame to capture the dynamic information of the speech signal. The super MFCCs frames can be combined into a 2-D MFCCs features map (MFCCsmap). Finally, the speaker model can be built based on the analysis sparse model and the sparse representations of the MFCCsmap are used as the LTA features. A state-of-the-art deep neural network (DNN) is employed as a classifier for speaker recognition. The experimental results illustrate the effectiveness and robustness of the proposed system.

Improving the accuracy of wind speed forecast can reduce the randomness and uncertainty of the wind power output and effectively improve a system's wind power accommodation. However, the high-dimensional historical wind speed information should be taken into account in the wind speed forecast, which increases the complexity of the model and reduces the efficiency and accuracy of a forecast. Feature selection by the Filter method can effectively reduce the feature dimension, but losing all the information of low-importance features. Although the feature reduction can retain the partial information of all features, it causes the loss of the partial information of high-importance features. In order to reduce the information loss caused by traditional FS and FR, short-term wind speed forecast with low information loss based on OSVD feature generation is proposed. First, the original wind speed series is denoised by OVMD. Then, based on the 96-dimensional original wind speed feature set, the OSVD is used to generate features. Furthermore, the extended original feature set EFS is obtained by combining the initial feature set with the features generated by OSVD. Gini importance is used to measure the importance of all features in EFS, and the forward feature selection is combined with random forests to determine the optimal subset. Finally, the optimal model determined by the new method is compared with seven models to verify the advancement of the new method. The experiments show that it reduces the information loss. Thus, the model has a higher forecast accuracy than the traditional model.

Finding out the wind deviation fault of the insulators of transmission lines in time can not only help people better complete the wind bias warning and prevention work, but also reduce the economic loss and the maintenance difficulty. A technology of Actual Wind Deviation Monitoring for Suspension Insulator Strings Based on Improved Edge Detection is proposed in this paper. It can calculate the actual wind deviation angle and distance of the insulator through image processing technology and camera calibration. Firstly, the video monitoring device is installed on the tower to collect the field insulator images, and the R + G improved grayscale processing and median filtering are carried out. Secondly, two-dimensional Otsu threshold segmentation is applied to the pre-processed image, and the insulator string target is obtained by combining morphological filtering and connection domain extraction. Then, the improved Kirsch operator is used to obtain the complete single-pixel wide edge of the insulator string. Searching the left and right boundary points of each insulator separately and the center points of each insulator can be calculated. Finally, these center points are transformed into world coordinates by camera calibration and fitted by least square method (LSM). The deviation angle θ of the insulator string can be obtained according to the slope of the fitting line, and the deviation distance d according to the length of the insulator string. When θ is greater than 7.5° and d is greater than 300mm, an alarm is issued to alert the staff. We analyze the performance of the technology by a series of experiments, the maximum error between the proposed method and the manual method is 8.72% and the minimum is 1.01%. Moreover, by analyzing 200 field insulator images, the identification accuracy are 93.75% and 91.36% respectively, which shows that the proposed method is effective and practical. It provides a new idea for the wind deviation monitoring of suspension insulator strings.

A novel real-valued algorithm of the popular root multiple signal classification of non-circular sources (NC-root-MUSIC) is developed for low-complexity direction of arrival (DOA) estimation. The proposed real-valued NC-root-MUSIC (RV-NC-root-MUSIC) reduces the computational burden mainly in following three aspects. First, unlike the classical real-valued algorithms, such as unitary transform, the complex-valued operation is still implemented when acquiring DOAs. However, the proposed RV-NC-root-MUSIC implements real-valued received data. Therefore, both tasks of eigenvalue decomposition (EVD) (or singular value decomposition (SVD)) and polynomial rooting can be implemented completely by real-valued computations in the proposed RV-NC-root-MUSIC. Next, due to the conjugate expansion of the non-circular sources, the steering vector dimension is extended by a factor of two, and the real-valued signal- and noise-subspace can be calculated by two reduced-dimension sub-matrices. As such, large-scale EVD/SVD computation is avoided. Finally, a real coefficient rooting polynomial with a similar order as NC-root-MUSIC is constructed. A variable substitution technique is further proposed to reduce the degree of this polynomial by half without sacrificing accuracy, which allows fast coefficient computation and real-valued rooting. The real-valued computation and reduced-order polynomial rooting in the proposed method result in a significant reduction in computational cost as compared to state-of-the-art methods. Numerical computer simulations illustrate that with significantly reduced complexity, the fast RV-NC-root-MUSIC algorithm can provide similar root mean square errors (RMSE) close to the conventional methods.

Icicle will cause serious distortion of insulator electric field, weaken its voltage strength, and easily cause insulator flashover. Previous studies have shown that the longer the length of icicles on insulator, the greater the possibility of electric flashover. Then, it is of great theoretical significance and practical value to study the image identification technology of icicle length of iced insulators. Firstly, the growth mechanism and structural characteristics of insulator icicles are discussed in this paper, and the image acquisition method of iced insulator is discussed. Secondly, the methods of saliency detection and color feature analysis are used to extract the region of icicle and insulator sheds respectively, then, the icicle structure on insulator is obtained by the difference calculation of above two images. Meanwhile, the pixel curve of the iced insulator image is processed by Fourier transform, and the reference size of insulators icicle is obtained by calculating the length of the shed spacing. Finally, a practical case about iced insulators in substations is analyzed in detail, which proves the effectiveness of the method proposed in this paper.

Translational compensation is one of the key problems in parameter estimation of moving targets and radar imaging, and envelope correction is the basis of translation compensation. However, in inverse synthetic aperture imaging, the traditional translation compensation algorithms cannot be applied to micro-motion targets. Based on the characteristics of micro-motion targets and the advantages of the terahertz radar, a new method of envelope correction for micro-motion targets based on the multi-layer perceptron is proposed in this paper, which is verified by a radar system with a carrier frequency of 330 GHz. The experimental targets adopted in this paper are rotating corner reflectors and precession warhead. Finally, this paper proposes a measure based on inverse Radon transform and compares the performance of the proposed algorithm with that of the previous one, which fully verifies the effectiveness of the proposed method.

In this paper, a modified belief propagation (BP) polar decoder is proposed. Unlike the orignal BP polar decoders, a check node is added to each node of the proposed decoder. In the BP decoding, the propagated messages of the nodes are modified by multiplying the messages from the check nodes, so as to enhance the reliability of these propagated messages. Numerical results show that the proposed decoder achieves better performance than that of the original BP decoders, only at cost of some additional multiplications, which indicates its effectiveness. [ABSTRACT FROM PUBLISHER]

With the ever-increasing amount and complexity of remote sensing image data, the development of large-scale image segmentation analysis algorithms has not kept pace with the need for methods that improve the final accuracy of object recognition. In particular, the development of such methods for large-scale images poses a considerable challenge. Traditional image segmentation methods are of high time complexity and low segmentation accuracy. This paper presents a new large-scale remote sensing image segmentation method that combines fuzzy region competition and the Gaussian mixture model. The new algorithm defines a non-similarity measurement for the class attribute of pixels using the Gaussian mixture model. This algorithm has the ability to perform high precision fitting of data with a statistical distribution, thereby effectively eliminating the influence of noise on the segmentation result. Moreover, fuzzy region competition is introduced to define the prior probability of the neighborhood that will be regarded as the weight of the Gaussian component. This process enhances the robustness of large-scale image segmentation. Finally, we acquire the image data from Google Earth and conduct experiments; the experimental results show that this new method has the feasibility and effectiveness and can achieve highly accurate segmentation results compared with current state-of-the-art methods.

The multilayer film consisting of MoO3/Ag/MoO3 (shortened as MAM) is very promising as an alternative of indium-tin oxide (ITO) to work as the transparent anode in organic solar cells (OSCs). In MAM-based thin OSCs, the absorption of light is quite poor. Here, we propose to apply a short-pitched metallic grating into the MAM-based OSC for improving the absorption in its thin active layer. Numerical calculations reveal that the obtained enhancement of integrated absorption in the active layer is more than 500% with respect to the equivalent planar control. The field distributions at the two peaks of the normalized absorption spectrum indicate that propagating surface plasmon polaritons (SPPs), as well as the horizontal metal-insulator-metal (MIM) waveguide mode, play the main role of trapping light into the active material. It is expected that the proposal could contribute to the development of the efficient ITO free thin OSCs.

In this paper, the improvement of the method measuring the junction temperature of lightemitting diodes (LEDs) has been studied experimentally. A practical method is proposed with only three measurement procedures. With the consideration of indium (In) composition and blue shift, the method has a high applicability, which is practical for the LED chips vary from blue to green chips under different currents, including the packaged chips. On the other hand, according to the experimental and derived results, the junction-temperature difference and peak-wavelength shift in both blue-shift and red-shift fields show similar parabolic-like relations. To simplify the experimental processes, dual-wavelength LEDs were fabricated and measured instead of conventional single-wavelength LEDs.

We have designed a patterned OSC based on 2-D arrayed dielectric nanospheres loaded on the indium tin oxide anode with other functional layers conformally embossed in the same profile as the nanosphere array. This design shows an integrated light absorption efficiency of 76.6% in the PSBTBT:PC71BM active layer over the wavelength range from 400 to 900 nm at normal incidence, outperforming the structurally identical planar control cell by 21.6%. Detailed investigations reveal that the excitation of photonic modes including the hybridization of cavity modes and Bloch modes, plasmonic modes, and their mutual coupling are responsible for the observed broadband enhancement in light absorption. Among them, the photonic modes are confirmed to be main contributor at most of the absorption band of PSBTBT:PC71BM, whereas at the absorption band edge of PSBTBT:PC71BM, the plasmonic resonance joins with the photonic resonances, collaboratively producing an enhancement factor as high as 110%. Moreover, the broadband absorption enhancement of our proposal is insensitive to the incident angle, favoring the practical application. This work provides a low cost and efficient route for achieving high-performance OSCs in experiment.

Plasmonic nanostructures have a great potential for enhancing light absorption of organic solar cells (OSCs). Our previous work has demonstrated that light absorption for OSCs with thin active layer can be significantly increased with a 1-D short-pitched metal grating, but the absorption enhancement is sensitive to the polarization of light and vanishes when the active layer is thicker than 60 nm. In this work, we extend the grating into 2-D space, specifically, with metallic nanocylinders packed in a hexagonal array embossed at the silver cathode. Numerical calculations indicate that the proposed OSC not only possesses polarization insensitivity but outperforms the corresponding equivalent planar device as well when the thickness of the active layer changes over a wide range from 40 to 120 nm. The absorption enhancement factor increases with the decrease of the active layer thickness, reaching 81.5% when the active layer is 40-nm thick. The evident enhancement in absorption is mainly due to the excitation of the strong dipole-like surface plasmon resonance, as well as their mutual coupling between neighboring nanocylinders. The present work could provide a promising route for the development of high-efficiency OSCs.