Search

Broad Learning System (BLS) is an emerging network paradigm that has received considerable attention in the regression and classification fields. However, there are two deficiencies which seriously hinder its deployment in real applications. The first one is the internal correlations among samples are not fully considered in the modeling process. Second, the strict binary label matrix utilized in BLS provides little freedom for classification. In this paper, to address the above issues, we propose to impose group-sparsity constraints on the class-specific transformed features and label error terms, respectively. The effect is not only the more appropriate margins between data can be preserved, but also the learnt label space can be flexible for recognition. As a result, the obtained projection matrix can show more vital discriminative ability. Further, we employ the alternating direction method of multipliers to solve the resulting optimization problem. Extensive experiments and analysis on diverse benchmark databases are carried out to confirm our proposed model’s superiority in comparison with other competing classification methods.

Broad Learning System (BLS) has been proven to be one of the most important techniques for classification and regression in machine learning and data mining. BLS directly collects all the features from feature and enhancement nodes as input of the output layer, which neglects vast amounts of redundant information. It usually leads to be inefficient and overfitting. To resolve this issue, we propose sparse regularization-based compact broad learning system (CBLS) framework, which can simultaneously remove redundant nodes and weights. To be more specific, we use group sparse regularization based on [Formula: see text] norm to promote the competition between different nodes and then remove redundant nodes, and a class of nonconvex sparsity regularization to promote the competition between different weights and then remove redundant weights. To optimize the resulting problem of the proposed CBLS, we exploit an efficient alternative optimization algorithm based on proximal gradient method together with computational complexity. Finally, extensive experiments on the classification task are conducted on public benchmark datasets to verify the effectiveness and superiority of the proposed CBLS.

To address the "overdetection" and "overtracking" problems experienced by traditional vehicle counting methods at low frame intervals, this paper proposes a vehicle counting method based on attention mechanism single-shot object detection (SSD) and state detection. First, we designed an attention-based saliency enhancement (ASE) module to fuse different scales of attention-enhanced feature mapping. Second, ASE modules were integrated into the SSD layers (from conv11 to conv15_2) to construct a vehicle detector for multiscale targets, named SSD-ASE. SSD-ASE was then used to detect and filter out vehicles located far from the identification line, where a filtering mechanism based on state detection that divides vehicles into four states based on their relative positions and the position identification line was used. Finally, by using the proposed state-based detector, the states of a vehicle were tracked using a relevant filtering algorithm, and counting was performed according to the state changes. The results of experiments performed on real traffic images/videos that we collected and on those obtained from the GRAM-RTM dataset showed that the proposed method achieved a higher detection accuracy (mAP = 95.94%) than the state-of-the-art methods and a higher tracking accuracy than traditional counting methods, while reducing the tracking number by approximately 60%.

Grain storage plays a vital part in social stability as well as people’s daily lives. By virtue of its unique storage advantages, wheat has become one of the most important grain reserves in the world. Wheat kernels continue consuming their own energy to maintain lives during the whole process of storage, which will lead to deterioration of their edible qualities as well as may be infected by certain mold provided that the surrounding conditions are suitable. Thus, establishing an efficient and accurate method to assess wheat kernel’s storage status is necessary for the sake of shaping a shelter to human beings as well as animals. For that reason, various wheat aging and wheat moldy detection methods have been summarized in this chapter. Up to now, study of aging and mold detection in wheat kernels can be divided into conventional chemical techniques and newly proposed physical detection methods. This review has analyzed in detail the limitations of conventional technologies, and then probes into some promising physical technologies for better detecting stored status of wheat kernels.

The quantity security of stored grain in warehouses is crucial for a country. Since manual inspection of stored grain quantity is inefficient, some online automeasurement methods have been proposed recently, but they are either time-consuming or too expensive to be popularized. This article proposes a deep learning-based method to automatically monitor changes in grain quantity of granaries. First, the image of the same scene in a granary is taken at different moments using a device that integrates a camera and an infrared laser rangefinder. Then, a deep semantic segmentation model based on an encoder–decoder framework is developed to extract the grain loading line and grain surface of the image. Finally, the distance and area between the extracted grain loading line and grain surface are calculated and compared with the previous measured one to determine whether the grain quantity of the granary has changed. To improve the accuracy of the segmentation results, a novel reverse attention model is proposed to provide guidance information to fuse low-level features, which calculates multiscale attention maps based only on the output of the last layer of the encoder. Furthermore, the proposed method and infrared laser rangefinder-based method are combined to get more accurate grain volume after an abnormal change of grain quantity is detected by our method. Experimental results show that our method is effective and feasible for monitoring changes in grain quantity and outperforms the state-of-the-art methods on semantic segmentation.

Surface defect classification (SDC) is the visual inspection of the surface of an object to identify appearance defects. Efficient and accurate SDC is i mportant for improving the quality of industrial products. A manually designed convolutional neural network (CNN) is traditionally used for SDC. In this study, a simpler SDC scheme with a higher classification accuracy, named NAS-SDC, is developed based on the neural architecture search (NAS) technique. A max-pooling cell based on NASNet is introduced to reduce the search space and the number of network parameters, thus simplifying the candidate operators for the search. Two network architectures are proposed to stack the search candidates or the best cells. The proposed method can be used to automatically design an efficient CNN model for SDC on a specific dataset. Experimental results show that the proposed method can find the best cells in ~11 h using a single graphics processing unit (GPU) and achieves higher classification accuracies (99.98%, 99.8% and 99.26%) than state-of-the-art methods on the Northeastern University (NEU-CLS), DAGM, and bridge defect datasets. The number of network parameters used in the proposed method is only 0.35 M, and the average test time per sample is approximately 61 ms, thus achieving a balance between performance and speed.

Regularly checking the quantity of stored grain in warehouses is essential for the grain safety of a country. However, current manual inspection ways fail to get real-time measurement results and require spending a lot of manpower and resources. In this paper, we proposed a computer vision-based method to automatically monitor the change in grain quantity of a granary. The proposed method was motivated from the observation that warehouse managers can use a camera to remotely monitor the grain security of a granary, which determines whether grain quantity is reduced by checking the distance between the grain surface and the grain loading line at the outlet of a granary. To this end, images were first captured by a camera, and a two-level spatial constraints-based SVM classifier was learned to detect the grain surface and the grain loading line of the images. During the test phase, the detected result of a test image obtained by SVM was further refined by GrabCut with higher order potentials to get the more accurate segmentation result. Finally, the area between the grain surface and the grain loading line was calculated, and then compared with the previous measured one to determine whether the grain surface had dropped. The experiment results validate the effectiveness of the two-level spatial constraints SVM and the strategy for monitoring the change in grain quantity.

Voltage-dependent resistors (VDRs) are important circuit-protection devices. Their performance is affected by packaging quality. To identify VDR packaging defects more accurately and efficiently, we have proposed a convolutional neural network (CNN)-based VDR appearance quality inspection method that includes four stages: image acquisition, data augmentation, neural architecture design, and CNN training and testing. In designing the neural architecture, we have proposed two VDR-oriented network blocks, which consist of a compressed subnet and a multiscale subnet. Then, a stacking-block-based neural architecture design (BlockNAD) strategy is employed to determine the number of blocks. The last block is connected to a classification layer composed of a global average pooling (GAP) layer and a full connection (FC) layer. Further, using a VDR dataset containing 8058 images, we compared the identification performances of the candidate networks with different structures on 12 categories of VDR defects by adopting a variety of indicators, such as the mean average precision (mAP) and average test time per sample. The experimental results of the proposed method demonstrate competitive results compared to the state-of-the-art methods in identifying VDR defects, with a mAP value of approximately 99.9% and an average test time per sample of approximately 3 ms.

The vibration and noise produced by the powertrain and waves inside ship cabins limit working efficiency and crew and passengers’ accommodation quality. This paper simplifies ship cabins as cavities and explores active control techniques to attenuate sound transmission via multiple parallel-supported flexible subplates. The theoretical formulations of the interaction between multiple subplates and cavities were performed and the coupling relationships between them were analyzed. Based on the multiple subplates and the cavity coupling models, numerical simulations were performed using the derived optimal controller to minimize the transmission of sound into the cavities through two and nine parallel-supported subplates. The various control strategies were explored to minimize the coupling system’s acoustic potential energy, and the control performances were compared and discussed. The mechanism of reducing sound transmission through multiple supported subplates into a cavity is revealed. The simulation results showed that the vibration pattern of the controlled subplate is changed after it is regulated, which increases its radiation to subdue the other subplates’ radiation, while increasing vibration of the controlled subplate. The more subplates a cavity has, the more kinetic energy the controlled subplate possess. Furthermore, the noise reduction performance of a cavity with fewer subplates is better than that with more subplates.

Made available in DSpace on 2020-12-10T17:06:00Z (GMT). No. of bitstreams: 0 Previous issue date: 2020-01-14 Operational modal analysis is an experimental modal analysis approach, which uses vibration data collected when the structure is under operating conditions. Amongst the methods for operational modal analysis, blind source separation-based methods have been shown to be efficient and powerful. The existing blind source separation modal identification methods, however, require the number of sensors to be at least equal to the number of modes in the frequency range of interest to avoid spatial aliasing. In this article, a frequency domain algorithm that overcomes this problem is proposed, which is based on the joint diagonalization of a set of weighted covariance matrices. In the proposed approach, the frequency range of interest is partitioned into several frequency ranges in which the number of active modes in each band is less than the number of sensors. Numerical simulations and an experimental example demonstrate the efficacy of the method. Harbin Engn Univ, Harbin, Peoples R China Univ Lyon, INSA Lyon, Lyon, France Univ Estadual Paulista, Sao Paulo, Brazil Univ Estadual Paulista, Sao Paulo, Brazil

Surface defect detection is an important way to improve the production quality of voltage-dependent resistors (VDRs). To improve the accuracy and efficiency of VDR surface quality detection, an end-to-end surface quality detection method based on deep convolutional neural networks (CNNs) was proposed. The method includes four stages: data preparation, convolution neural network design, CNN training, and testing. First, images of VDRs were acquired from three perspectives, i.e., the front, back, and side, and then training, validation and testing sets were obtained. Second, the proposed CNN models for VDR surface defect detection were constructed. Third, during the training stage, the images with class labels from the established training sets were input to the proposed network for training and validation. Finally, in the testing stage, test images from a total of 408 samples of two VDR models were used to test the trained network. The sensitivity, specificity, accuracy, precision and F measure of the proposed algorithm were compared with those of state-of-the-art methods, and the experimental results showed that the proposed method has a high recognition speed and accuracy and meets the requirements of online real-time detection.

When exposed to high levels of noise, earmuffs are often used to avoid hearing loss. However, active noise control earmuffs may exhibit nonlinearities under excessive levels of noise, due to their low-power characteristics of the loudspeakers, and thus nonlinear control algorithms are required to improve the control performance. In this paper, an analytical model of a nonlinear active noise control earmuff is investigated. Based on this model, a robust state feedback control law is designed in the framework of linear matrix inequalities with respect to the parametric uncertainties of the loudspeaker and the limitation of control input. Then the backstepping approach is adopted to force the nonlinear part of the loudspeaker to track the derived state feedback signal and estimate the unknown parameters. Both recorded vehicle noise and multi-frequency noise are used to test the effectiveness of the proposed controller and the control performance is compared with that of a widely accepted nonlinear generalized functional link artificial neural network algorithm. Simulation results demonstrate that the proposed controller is capable of attenuating the interior noise and reducing harmonic and intermodulation distortions significantly.

The segmentation of brain tumors in magnetic resonance imaging (MRI) images plays an important role in early diagnosis, treatment planning and outcome evaluation. However, due to gliomas’ significant diversity in structure, the segmentation accuracy is low. In this paper, an automatic segmentation method integrating the small kernels two-path convolutional neural network (SK-TPCNN) and random forests (RF) is proposed, the feature extraction ability of SK-TPCNN and the joint optimization capability of model are presented respectively. The SK-TPCNN structure combining the small convolutional kernels and large convolutional kernels can enhance the nonlinear mapping ability and avoid over-fitting, the multiformity of features is also increased. The learned features from SK-TPCNN are then applied to the RF classifier to implement the joint optimization. RF classifier effectively integrates redundancy features and classify each MRI image voxel into normal brain tissues and different parts of tumor. The proposed algorithm is validated and evaluated in the Brain Tumor Segmentation Challenge (Brats) 2015 challenge Training dataset and the better performance is achieved.

Nonrigid registration of medical images is especially critical in clinical treatment. Mutual information is a popular similarity measure for medical image registration; however, only the intensity statistical characteristics of the global consistency of image are considered in MI, and the spatial information is ignored. In this paper, a novel intensity‐based similarity measure combining normalized mutual information with spatial information for nonrigid medical image registration is proposed. The different parameters of Gaussian filtering are defined according to the regional variance, the adaptive Gaussian filtering is introduced into the local structure tensor. Then, the obtained adaptive local structure tensor is used to extract the spatial information and define the weighting function. Finally, normalized mutual information is distributed to each pixel, and the discrete normalized mutual information is multiplied with a weighting term to obtain a new measure. The novel measure fully considers the spatial information of the image neighborhood, gives the location of the strong spatial information a larger weight, and the registration of the strong gradient regions has a priority over the small gradient regions. The simulated brain image with single‐modality and multimodality are used for registration validation experiments. The results show that the new similarity measure improves the registration accuracy and robustness compared with the classical registration algorithm, reduces the risk of falling into local extremes during the registration process.

Melanoma is the deadliest form of skin cancer, and its incidence is increasing. The first step in automated melanoma analysis of dermoscopy images is to segment the area of the lesion from the surrounding skin. To improve the accuracy and adaptability of segmentation, an algorithm called sampling with level set by integrating color and texture (SLS-CT) is proposed that not only designs a new way to incorporate textural and color features in the definition of the energy functional but also utilizes an index called texture level, proposed in this work, to automatically decide the weight of each feature in the combined energies. First, at the preprocessing stage, hair and black frame removal is applied, and a potential lesion area is then obtained using Otsu thresholding and entropy maximization. Thereafter, the probability distribution of prior color in this potential lesion area is calculated as well. Second, Gabor wavelet-based texture features are extracted and integrated with the prior color into the evolving energies of the level set based on the texture level. To achieve global optimization, a Markov chain Monte Carlo sampling approach guided by the combined energy is adopted in evolving the level set, which ultimately defines a border in the image to segment a lesion from normal skin. Finally, morphological operations are used for postprocessing. The experimental results of the proposed algorithm are compared with those of other state-of-the-art algorithms, demonstrating that the proposed algorithm outperforms the other tested ones in terms of accuracy and adaptability to different databases.

Clustering refers to the problem of partitioning data into several groups according to the predefined criterion. Graph-based method is one of main clustering approaches and has been shown impressive performance in many literatures. The core issue of graph-based clustering is how to construct a good adjacency graph. A large number of works employ the sparse representation of data as the similarity measure by ℓ1 regularization. However, due to the flat nature of the ℓ1 norm, such methods solve the sparse representation of each data point individually, which do not take into account the global structure of data. To exploit the global and essential structure in data, in contrast to existing methods, we propose to learn a graph with group sparsity. To incorporate more information into the graph, we also use the manifold regularization with adaptive similarity during the process of group sparse self-representation. The resulting model is handled by Alternating Direction Method of Multipliers (ADMM). Further, we employ Iterative Re-weighted Least Squares (IRLS) algorithm and threshold operator to solve the ADMM subproblems. Experimental results on real-world datasets demonstrate the superiority of our method compared to the competing clustering methods.

OBJECTIVE: In order to solve the blurred structural details and over-smoothing effects in sparse representation dictionary learning reconstruction algorithm, this study aims to test sparse angle CT reconstruction with weighted dictionary learning algorithm based on adaptive Group-Sparsity Regularization (AGSR-SART). METHODS: First, a new similarity measure is defined in which Covariance is introduced into Euclidean distance, Non-local image patches are adaptively divided into groups of different sizes as the basic unit of sparse representation. Second, the weight factor of the regular constraint terms is designed through the residuals represented by the dictionary, so that the algorithm takes different smoothing effects on different regions of the image during the iterative process. The sparse reconstructed image is modified according to the difference between the estimated value and the intermediate image. Last, The SBI (Split Bregman Iteration) iterative algorithm is used to solve the objective function. An abdominal image, a pelvic image and a thoracic image are employed to evaluate performance of the proposed method. RESULTS: In terms of quantitative evaluations, experimental results show that new algorithm yields PSNR of 48.20, the maximum SSIM of 99.06% and the minimum MAE of 0.0028. CONCLUSIONS: This study demonstrates that new algorithm can better preserve structural details in reconstructed CT images. It eliminates the effect of excessive smoothing in sparse angle reconstruction, enhances the sparseness and non-local self-similarity of the image, and thus it is superior to several existing reconstruction algorithms.

Background Brain tumor segmentation plays an important role in assisting diagnosis of disease, treatment plan planning, and surgical navigation. Objective This study aims to improve the accuracy of tumor boundary segmentation using the multi-scale U-Net network. Methods In this study, a novel U-Net with dilated convolution (DCU-Net) structure is proposed for brain tumor segmentation based on the classic U-Net structure. First, the MR brain tumor images are pre-processed to alleviate the class imbalance problem by reducing the input of the background pixels. Then, the multi-scale spatial pyramid pooling is used to replace the max pooling at the end of the down-sampling path. It can expand the feature receptive field while maintaining image resolution. Finally, a dilated convolution residual block is combined to improve the skip connections in the training networks to improve the network's ability to recognize the tumor details. Results The proposed model has been evaluated using the Brain Tumor Segmentation (BRATS) 2018 Challenge training dataset and achieved the dice similarity coefficients (DSC) score of 0.91, 0.78 and 0.83 for whole tumor, core tumor and enhancing tumor segmentation, respectively. Conclusions The experiment results indicate that the proposed model yields a promising performance in automated brain tumor segmentation.

Made available in DSpace on 2020-12-12T01:25:02Z (GMT). No. of bitstreams: 0 Previous issue date: 2020-05-01 An active vibration isolation system is retrofitted to a diesel generator set in a tugboat to determine the effectiveness of such a system in a realistic practical environment. The system consists of six bespoke inertial actuators chosen to make minimal modifications to the machinery arrangement, and a DSP-based controller. Six accelerometers are collocated with the actuators on the top of six isolators to act as error sensors, and six accelerometers are placed below the isolators to give a measure of the global vibration of the ships structure below the generator set. A hydrophone is also placed in the water to give an indication of the underwater noise due to the generator. The control strategy employed is six-input and six-output decentralized adaptive feedforward control with the reference signal being derived from the signal from an optical tachometer on shaft between the engine and the generator. To suppress the vibration at all the dominant forcing frequencies, an electrical circuit generated the half engine orders required from the measured reference signal. The experimental results show that the combination of the active control system and the passive isolators is effective in reducing the global vibration and the acoustic pressure at the hydrophone position. Institute of Vibration and Noise Control Harbin Engineering University Departamento de Engenharia Mecânica Universidade Estadual Paulista (UNESP) Departamento de Engenharia Mecânica Universidade Estadual Paulista (UNESP)

Since morphology of retinal blood vessels plays a key role in ophthalmological disease diagnosis, retinal vessel segmentation is an indispensable step for the screening and diagnosis of retinal diseases with fundus images. In this paper, deep convolution adversarial network combined with short connection and dense block is proposed to separate blood vessels from fundus image, named SUD-GAN. The generator adopts U-shape encode-decode structure and adds short connection block between convolution layers to prevent gradient dispersion caused by deep convolution network. The discriminator is all composed of convolution block, and dense connection structure is added to the middle part of the convolution network to strengthen the spread of features and enhance the network discrimination ability. The proposed method is evaluated on two publicly available databases, the DRIVE and STARE. The results show that the proposed method outperforms the state-of-the-art performance in sensitivity and specificity, which were 0.8340 and 0.9820, and 0.8334 and 0.9897 respectively on DRIVE and STARE, and can detect more tiny vessels and locate the edge of blood vessels more accurately. ELECTRONIC SUPPLEMENTARY MATERIAL: The online version of this article (10.1007/s10278-020-00339-9) contains supplementary material, which is available to authorized users.

As one of the important dimensionality reduction techniques, unsupervised feature selection (UFS) has enjoyed amounts of popularity over the last few decades, which can not only improve learning performance, but also enhance interpretability and reduce computational costs. The existing UFS methods often model the data in the original feature space, which cannot fully exploit the discriminative information. In this paper, to address this issue, we investigate how to strengthen the relationship between UFS and the feature subspace, so as to select relevant features more straightforwardly and effectively. Methodologically, a novel UFS approach, referred to as Graph Regularized Local Linear Embedding (GLLE), is proposed by integrating local linear embedding (LLE) and manifold regularization constrained in feature subspace into a unified framework. To be more specific, we explicitly define a feature selection matrix composed of 0 and 1, which can realize the process of UFS. For the purpose of modelling the feature selection matrix, we propose to preserve the local linear reconstruction relationship among neighboring data points in the feature subspace, which corresponds to LLE constrained in the feature subspace. To make the feature selection matrix more accurate, we propose to use manifold regularization as an assistant of LLE to find the relevant and representative features such that the selected features can make each sample under the feature subspace be accordance with the manifold assumption. A tailored iterative algorithm based on Alternative Direction Method of Multipliers (ADMM) is designed to solve the proposed optimization problem. Extensive experiments on twelve real-world benchmark datasets are conducted, and the more promising results are achieved compared with the state-of-the-arts approaches.

BACKGROUNDAccurate segmentation of brain tumor depicting on magnetic resonance imaging (MRI) is an important step for doctors to determine optimal treatment plan of Gliomas, which are the common malignant brain tumors that seriously damage patients' health and life.OBJECTThis study aims to improve accuracy and efficiency of brain tumor segmentation on MRI using the advanced deep learning model.METHODIn this study, an improved model based on the U-net for accurate segmentation of brain tumor MRI images, called Deeper ResU-net, is proposed. First, a deep Deeper U-net is built, which has deeper network depth compared with U-net, uses Squeeze Operator to control network parameters and attempts to enhance the feature extraction ability. Then, Deeper ResU-net is formed to eliminate degradation phenomenon of the deep network, in which residual unit is designed and integrated into the Deeper U-net to keep the number of parameters unchanged.RESULTDeeper ResU-net makes the deep network conduct stable training without degrading. Evaluation result shows that the Deeper ResU-net has achieved competitive result with average DSC metrics of 0.9, 0.82, 0.88 for Complete tumor region, Core tumor region and Enhanced tumor region, respectively.CONCLUSIONBy extending the U-net model to a deeper layer and adding the residual structure to ensure effective and stable training of the model, the experiment results demonstrate that applying the improved Deeper ResU-net can effectively eliminate the degradation phenomenon of deep network and improve segmentation performance.

This paper investigates the active noise control algorithms and improves them by using online secondary path modeling. The proposed method uses three adaptive filters to track the convergence of the system as well as reduce the target noise. By theoretical analysis, the optimized step size and injected random noise gain are derived. The step size is varied according to the convergence of three adaptive filters and the gain of injected random noise is proportional to the power of modeling error, which makes the method more stable even in the presence of strong perturbation. Compared with previous methods, the proposed method improves the convergence rate and estimation accuracy for both the active control system and the secondary path modeling process with less increase of computational complexity. The simulation results verify the above analysis by controlling three different kinds of noise.

The existing spectrum index-based methods for detecting vegetation coverage suffer from an over-dependence on spectrum. To address these issues, this paper proposes a graph cut-based variational level set segmentation algorithm that combines multi-channel local wavelet texture (MCLWT) and color. First, the prior color is generated by automatic estimation based on the mathematical morphology with a color histogram. Then, local wavelet texture features are extracted using a multi-scale and orientation Gabor wavelet transformation followed by local median and entropy filtering. Next, in addition to the energy of color, that of MCLWT is integrated into the variational level set model based on kernel density estimation. Consequently, all energies are integrated into the graph cut-based variational level set model. Finally, the proposed energy functional is made convex to obtain a global optimal solution, and a primal-dual algorithm with global relabeling is adopted to accelerate the evolution of the level sets. A comparison of the segmentation results from our proposed algorithm and other state-of-the-art algorithms showed that our algorithm effectively reduces the over-dependence on color and yields more accurate results in detecting vegetation coverage.

For acoustic beamforming, the Fast Iterative shrinkage-thresholding algorithm (FISTA) is efficient in improving the spatial resolution and the dynamic range of conventional beamforming maps. However, there are some drawbacks inherent to FISTA. In this paper, to further improve the convergence rate, numerical stability, and computation time of FISTA, adaptive step-size FISTA (AFISTA), Greedy FISTA (GFISTA) along with compression computation grid method (CG) are applied in the solving process of deconvolution beamforming. Two incoherent sinusoid sound sources with different distances, locations and frequencies are simulated and experimented to investigate the behaviors of these new algorithms (denoted as FISTA-CG, AFISTA-CG, and GFISTA-CG). Results show that all algorithms can accurately reveal the locations and amplitudes of sound sources with little differences in the high iterations. Although AFISTA-CG consumes more time during each iteration, it achieves the fastest convergence rate among these algorithms in the low iterations, and thus visualizes the accurate locations of sound sources the most rapidly in all cases. Furthermore, both AFISTA-CG and GFISTA-CG have lower convergence errors and numerical fluctuations compared with FISTA-CG.

Surface defect recognition is one of the key technologies for varistor quality inspection, which can greatly improve detection efficiency and performance. In order to more accurately identify the surface defects of a varistor body and the pins, a method for identifying the surface defects based on deep convolutional neural networks (CNN) is proposed. The proposed method mainly includes four stages: image acquisition and data set construction, convolutional neural network modeling, CNN training and testing. Firstly, varistor images are acquired, and the body and pins of the varistor are segmented by image segmentation method. The number of samples is increased by data augmentation to make a data set of 5 classes. Secondly, according to the appearance characteristics of varistor, a CNN model is designed for varistor surface defect recognition. Third, using the created data set, the training data set with category labels are input to the proposed CNN for training. Finally, 1200 test samples were tested on the trained model in the test phase and the performance of the proposed algorithm was evaluated using mean average precision. The experimental results show that our method can identify the surface defects of the main body and pins of varistor efficiently and accurately.

The edge of the traditional non-local Means (NLM) constrained CT reconstruction algorithm tends to be over-smooth. An improved algebraic iterative reconstruction algorithm based on adaptive non-local mean constraint (ART_SNLM) is proposed in this paper. Firstly, a clock-like similarity window shape is defined, the pixels with high similarity can be selected to participate in the weight calculation with higher probability. Secondly, in order to remove noise and protect the edges simultaneously, the adaptive filter parameters are designed to filter the reconstructed image through the change of the difference between the gray value of the neighborhood pixel and the center pixel in the clock-like window's three directions, and parameter is related to the change of the number of iterations as well. The improved algorithm is used to reconstruct the classic Shepp-Logan phantom.The experimental results show that the reconstructed image by ART_SNLM algorithm is not only closer to the real phantom, but also has a smaller reconstruction error, which can protect the edge characteristics of the image better.

Made available in DSpace on 2019-10-06T15:48:57Z (GMT). No. of bitstreams: 0 Previous issue date: 2019-06-01 This paper describes an analytical investigation into synchrophasing, a vibration control strategy on a machinery installation in which two rotational machines are attached to a beam-like raft by discrete resilient isolators. Forces and moments introduced by sources are considered, which effectively represent a practical engineering system. Adjusting the relative phase angle between the machines has been theoretically demonstrated to greatly reduce the cost function, which is defined as the sum of velocity squares of attaching points on the raft at each frequency of interest. The effect of the position of the machine is also investigated. Results show that altering the position of the secondary source may cause a slight change to the mode shape of the composite system and therefore change the optimum phase between the two machines. Although the analysis is based on a one-dimensional Euler–Bernoulli beam and each machine is considered as a rigid-body, a key principle can be derived from the results. However, the factors that can influence the synchrophasing control performance would become coupled and highly complicated. This condition has to be considered in practice. College of Power and Energy Engineering Harbin Engineering University Departamento de Engenharia Mecânica UNESP Departamento de Engenharia Mecânica UNESP

Brain tumor images segmentation plays a crucial role in the auxiliary diagnosis of disease, treatment planning and surgical navigation. In order to accurately segment brain tumor images, this paper proposes an automatic brain tumor Magnetic Resonance Imaging (MRI) image segmentation algorithm based on the U-net model. Firstly, based on samples preprocessing, loss function and optimization algorithm, a U-net baseline model with optimal parameters structure which is suitable for segmentation task is constructed. Secondly, a feature recombination layer, i.e. a 1×1 convolutional layer, is added to the baseline model to linearly recombine the upper layer features, enrich extracted features, reduce network parameters, and improve the segmentation results. The algorithm is validated and evaluated on the Brain Tumor Segmentation Challenge (Brats) 2015 challenge training dataset. The experimental results show that the proposed algorithm has strong competitiveness compared with the existing brain tumor MRI image segmentation algorithm.

Medical image registration technology is important in medical image processing, which can establish correspondence between complex medical images and ensure the comparability of image data. Since the deep convolutional neural network model is prone to degradation and the convolution kernel has limited detection range during training, a Multi-scale residual full convolutional network (MS-ResFCN) model is proposed for unsupervised non-rigid medical image registration tasks. The model introduces a residual structure in the Fully Convolutional Network (FCN) model to ensure effective and stable training. At the same time, a hierarchical multi-scale convolution kernel is constructed within a single convolutional layer of the residual structure, which enhances the nonlinear mapping ability of the network. The local texture information and the contextual information of the image are synchronously extracted and fused to enrich the diversity of features. The experiment results on LPBA40 data sets show that the MS-ResFCN model can effectively eliminate the degradation phenomenon of the deep network during the training process and extract multi-scale features, which achieves better feature representation ability and registration accuracy.

This paper investigates active broadband noise control inside vehicles with a multichannel controller. The noncausal inversion of a practical nonminimum-phase secondary path is formulated, and its influence on noise-reduction performance is analyzed. Based on multiple coherence between reference signals and undesired noise, a novel formulation for identifying primary paths with correlated excitation signals is presented and a causal optimal controller is proposed. Meanwhile, the proposed controller can be used as an accurate predictor to estimate the maximal achievable noise reduction and provide a reference to improve the control systems. The robustness of the proposed algorithm is examined by varying the uncertainty of primary paths. Finally, the performance of the proposed causal optimal controller is validated using the data measured in a car. The results show that the proposed algorithm outperforms traditional algorithms and achieves a significant broadband noise reduction in time-invariant systems.

The compressed sensing (CS)-based methods for tomography synthetic aperture radar (SAR) imaging have a good performance in the case of a large number of baselines. Unfortunately, for the current tomography SAR, the baselines are obtained in repeat-pass mode from a large number of parallel passes in the same scene. That is expensive and can be severely affected by temporal decorrelation. And the performance of CS-based methods degrades rapidly with the decrease of the number of baselines. Inspired by the block-sparsity theory, an extended block orthogonal matching pursuit algorithm by using the weighted operation and a new column-block selection strategy is proposed for tomography SAR imaging. By using neighboring pixels information in reconstruction, the proposed method can improve the performance of CS-based methods in the case of strong noise and a small number of baselines. Experimental results confirm the effectiveness of the proposed method.

A novel design of a two-stage nonlinear vibration isolation system, with each stage having a high-static-low-dynamic stiffness (HSLDS), is studied experimentally in this paper. The positive stiffness in each stage is realized by a metallic plate, and the corresponding negative stiffness is realized by a bistable carbon fiber–metal (CF) composite plate. An analytical model is developed as an aid to design a bistable composite plate with the required negative stiffness, and a static test of the plate is conducted to measure the actual stiffness of the plate. Dynamic tests of the two-stage isolator are carried out to determine the effectiveness of the isolator. Two tests are conducted, one with the bistable composite plates removed so that the isolator behaves as a linear device and one with the bistable composite plates fitted. An improvement in the isolator transmissibility of about 13 dB at frequencies greater than about 100 Hz is achieved when the bistable composite plates are added.

Intelligent video surveillance is to analysis video or image sequences captured by a fixed or mobile surveillance camera, including moving object detection, segmentation and recognition. By using it, we can be notified immediately in an abnormal situation. Pedestrian detection plays an important role in an intelligent video surveillance system, and it is also a key technology in the field of intelligent vehicle. So pedestrian detection has very vital significance in traffic management optimization, security early warn and abnormal behavior detection. Generally, pedestrian detection can be summarized as: first to estimate moving areas; then to extract features of region of interest; finally to classify using a classifier. Redundant wavelet transform (RWT) overcomes the deficiency of shift variant of discrete wavelet transform, and it has better performance in motion estimation when compared to discrete wavelet transform. Addressing the problem of the detection of multi-pedestrian with different speed, we present an algorithm of pedestrian detection based on motion estimation using RWT, combining histogram of oriented gradients (HOG) and support vector machine (SVM). Firstly, three intensities of movement (IoM) are estimated using RWT and the corresponding areas are segmented. According to the different IoM, a region proposal (RP) is generated. Then, the features of a RP is extracted using HOG. Finally, the features are fed into a SVM trained by pedestrian databases and the final detection results are gained. Experiments show that the proposed algorithm can detect pedestrians accurately and efficiently.

Automated detection of blood vessel structures is becoming a crucial interest for better management of vascular disease. In this paper, we propose an algorithm for vessel segmentation in digital retinal images based on integral channel features and random forests. In the first stage, preprocessing is performed to obtain the candidate pixels of vessels, then a host of simple features are extracted for each candidate pixels based on integral channels. Furthermore random forests is used to classify the candidate pixels as vessels or not. Finally, postprocessing is applied to fill pixel gaps in classified blood vessels. The proposed algorithm achieves an average accuracy of 0.9614, 0.9588, sensitivity of 0.7191, 0.6996 and specificity of 0.9849, 0.9787 on two public databases DRIVE and STARE respectively.

Through utilizing the technology of compressive sensing (CS), the channel estimation methods can achieve the purpose of reducing pilots and improving spectrum efficiency. The channel estimation and pilot design scheme are explored during the correspondence under the help of block-structured CS in massive MIMO systems. The block coherence property of the aggregate system matrix can be minimized so that the pilot design scheme based on stochastic search is proposed. Moreover, the block sparsity adaptive matching pursuit (BSAMP) algorithm under the common sparsity model is proposed so that the channel estimation can be caught precisely. Simulation results are to be proved the proposed design algorithm with superimposed pilots design and the BSAMP algorithm can provide better channel estimation than existing methods.

An active control strategy with on-line secondary path modeling is proposed and applied in active control of helicopter structural vibration. Computer simulations of the secondary path modeling performance demonstrate the superiorities of the active control strategy. A 2-input 4-output active control simulation for a helicopter model is performed and greatreduction of structural vibration is achieved. 2-input 2-output and 2-input4-output experimental studies of structural vibration control for a free-free beam are also carried out in laboratory to simulate a flying helicopter. The experimental results also show better reduction of the structural vibration, which verifies that the proposed method is effective and practical in structural vibration reduction.

Subscriber radio location techniques for code division multiple access (CDMA) cellular networks has been studied extensively in recent years. The network-based angle of arrival (AOA), time difference of arrival (TDOA), and time of arrival (TOA) techniques offer solutions to the position estimation problem. In this paper, The signal processing scheme of wireless location system based IS-95 is presented, in which TOA of reverse access channel transmissions is measured using sub-correlation detection algorithms and TOA estimation accuracy is improved using second search. Furthermore, reverse access channel decoding is implemented used to identify access channel message type, mobile identification, and dialed number.© (2012) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.

In this paper, to solve the problem of feedback cost and performance loss for precoded spatial multiplexing with limited feedback, an novel nonlinear detection method based on limited feedback for precoded vblast is proposed, the corresponding criteria is used for select optimal codeword from the Grassmannian codebook at the receiving according to the channel state information, and feedback the precoding word index to the transmitter, thereby forming the equivalent channel. At the receiving, recycle and successive mapping from the entire space of constellation points as the first the underlying detection signals at the receiver, in order to maximize the performance of first detection layer, and then using the traditional detection algorithm for the remaining received signal, and determine a set of the optimal detection signal as decoding result based on the maximum likelihood criterion. Simulation results show that the proposed method to ensure system throughput, the cost of the feedback link and BER performance are improved some extent.

In Orthogonal Frequency Division Multiplexing (OFDM) system, Carrier frequency offsets (CFOs) between the transmitter and the receiver destroy the orthogonality between carriers and degrade the system performance significantly. On the basis of traditional frequency offset estimation algorithm, an improved carrier frequency offset estimation algorithm is presented in this paper, which uses one complex training sequence, can effectively estimate fractional frequency offset in the time domain and the integer frequency offset in the frequency domain. Simulation results show that the improved carrier frequency offset estimation algorithm performs better than conventional P. Moose and Schmidl algorithm, which can effectively improve the frequency estimation accuracy and provides a wide acquisition range for the carrier frequency offset with low complexity.

In recent publications, some special detectors were proposed to improve the performance and complexity gaps between the optimal detector and the suboptimal detector. In this paper, we propose a multistage chase detection algorithm for MIMO-OFDM systems, which use chase detection framework to design sub-detector, so the nest chase detection was implemented that can further decrease the error spread and improve the BER performance. Simulation results show that the proposed schemes performs better than BLAST and conventional chase detection, which can achieve near-ML performance with lower complexity compared with ML algorithm.

GPS-based twin antennas real-time attitude determination system realized with TMS320C6713 is presented in this paper. The direct computation method used to estimate attitude parameter is introduced firstly. Then the position, velocity and float ambiguity are estimated using kalman filter, the integer ambiguity is resolved based on Z-transformation augmented FASF algorithm. In the end, the software flow chart and hardware block diagram of attitude determination system were presented. The experiment results show that yaw accuracy about 0.04 degrees and pitch accuracy about 0.1 degrees at the output rate of 10Hz can be achieved.

Selecting hundreds of thousands of particles from low contrast and low SNR Cryo-electron micrograph (Cryo-EM) images is one of the major bottle-necks in advancing toward achieving atomic resolution reconstruction of biological macromolecules. To recognize true particles from impurities and other invalid particles, shape information plays a significant role. This paper provide a new method based on multi-resolution analysis to detect the main edges of objects in Cryo-EM image, which can be the pre-processing operations of next step of particle identification.

There still exist two key problems required to be solved in the classification rule extraction, i.e. how to select attributes and discretize continuous attributes effectively. The lack of efficient heuristic information is the fundamental reason that affects the performance of currently used approaches. In this paper, a new measure for determining the importance level of the attributes based on the trained SVM is proposed, which is suitable for both continuous attributes and discrete attributes. Based on this new measure, a new approach for rule extraction from trained SVM and classification problems with continuous attributes is proposed. The performance of the new approach is demonstrated by several computing cases. The experimental results prove that the approach proposed can improve the validity of the extracted rules remarkably compared with other rule extracting approaches, especially for the complicated classification problems.

This paper presents a general model of a two-stage vibration isolation system involving a flexible base plate with arbitrary boundary conditions. The dynamic behavior of the coupling system was obtained by using an impedance method in which the impedance matrix of the base plate was derived from a combined use of an improved Fourier series expansion and the Rayleigh-Ritz method. Subsequently, with the purpose of attenuation of sound radiation from the base plate structure, the optimal force information is formulated under various active control strategies including: 1) minimizing the total vibratory power of the coupled structure, 2) minimizing the sound power radiation from the base plate, 3) minimizing the vibratory power transmitted to the base plate, and 4) minimizing the mean square velocities at the isolator locations on the flexible plate. Numerical results were presented and discussed in detail. Finally, some concluding remarks are made.

In this study, the energy distributions and power flows between some basic structural components such as beams, plates, and shells are studied using a so-called Fourier Spectral Element Method (FSEM). Similar to the SEA modeling, a complex system is here also considered as an assembly of subsystems or components. The FSEM, however, is deterministic in nature in that the solution is obtained by directly and faithfully solving the governing equations for each component under the actual boundary and coupling conditions. What make this model powerful and unique lie in its capability and flexibility of effectively dealing with model uncertainties (due to the probabilistic/stochastic natures of some model parameters) and engineering and manufacturing errors which tend to become critically important at higher frequencies. Since this method does not involve any artificial assumptions or simplifications, it potentially offers a whole frequency solution with adaptive spatial and frequency resolutions.