Underwater Noise Target Recognition Based on Sparse Adversarial Co-Training Model with Vertical Line Array

The automatic identification of underwater noncooperative targets without label records remains an arduous task considering the marine noise interference and the shortage of labeled samples.In particular,the data-driven mechanism of deep learning cannot identify false samples,aggravating the difficulty in noncooperative underwater target recognition.A semi-supervised ensemble framework based on vertical line array fusion and the sparse adversarial co-training algorithm is proposed to identify noncooperative targets effectively.The sound field cross-correlation compression(SCC)feature is developed to reduce noise and computational redundancy.Starting from an incomplete dataset,a joint adversarial autoencoder is constructed to extract the sparse features with source depth sensitivity,aiming to discover the unknown underwater targets.The adversarial prediction label is converted to initialize the joint co-forest,whose evaluation function is optimized by introducing adaptive confidence.The experiments prove the strong denoising performance,low mean square error,and high separability of SCC features.Compared with several state-of-the-art approaches,the numerical results illustrate the superiorities of the proposed method due to feature compression,secondary recognition,and decision fusion.

Novel and Comprehensive Approach for the Feature Extraction and Recognition Method Based on ISAR Images of Ship Target

This paper proposes a novel and comprehensive method of automatic target recognition based on real ISAR images with the aim to recognize the non-cooperative ship targets. The special characteristics of the ISAR images for the real data compared with the simulated ISAR images are analyzed firstly. Then,the novel technique for the target recognition is proposed,and it consists of three steps,including the preprocessing,feature extraction and classification. Some segmentation and morphological methods are used in the preprocessing to obtain the clear target images. Then,six different features for the ISAR images are extracted.By estimating the features' conditional probability, the effectiveness and robustness of these features are demonstrated. Finally,Fisher's linear classifier is applied in the classification step. The results for the allfeature space are provided to illustrate the effectiveness of the proposed method.

Radar automatic target recognition based on feature extraction for complex HRRP

Radar high-resolution range profile （HRRP） has received intensive attention from the radar automatic target recognition （RATR） community. Usually, since the initial phase of a complex HRRP is strongly sensitive to target position variation, which is referred to as the initial phase sensitivity in this paper, only the amplitude information in the complex HRRP, called the real HRRP in this paper, is used for RATR, whereas the phase information is discarded. However, the remaining phase information except for initial phases in the complex HRRP also contains valuable target discriminant information. This paper proposes a novel feature extraction method for the complex HRRP. The extracted complex feature vector, referred to as the complex feature vector with difference phases, contains the difference phase information between range cells but no initial phase information in the complex HRRR According to the scattering center model, the physical mechanism of the proposed complex feature vector is similar to that of the real HRRP, except for reserving some phase information independent of the initial phase in the complex HRRP. The recognition algorithms, frame-template establishment methods and preprocessing methods used in the real HRRP-based RATR can also be applied to the proposed complex feature vector-based RATR. Moreover, the components in the complex feature vector with difference phases approximate to follow Gaussian distribution, which make it simple to perform the statistical recognition by such complex feature vector. The recognition experiments based on measured data show that the proposed complex feature vector can obtain better recognition performance than the real HRRP if only the cell interval parameters are properly selected.

A new feature extraction method using the amplitude fluctuation property of target HRRP for radar automatic target recognition

Due to the aspect sensitivity of high-resolution range profile(HRRP),traditional radar HRRP target recognition methods usually use average profile within some target-aspect region as the target-aspect template.Actually,the amplitude fluctuation property of target HRRP also represents some feature information of the target.Based on the scattering center model,a new feature extraction method using the amplitude fluctuation property of target HRRP is proposed in this paper.The weighted HRRP feature extracted by the new method can represent the scatterer distribution in every range cell,thereby it can describe the scattering property of the target better.The experimental results based on measured data show that the new feature extraction method can greatly improve recognition performances.

HRRP target recognition based on kernel joint discriminant analysis

With the improvement of radar resolution,the dimension of the high resolution range profile(HRRP)has increased.In order to solve the small sample problem caused by the increase of HRRP dimension,an algorithm based on kernel joint discriminant analysis(KJDA)is proposed.Compared with the traditional feature extraction methods,KJDA possesses stronger discriminative ability in the kernel feature space.K-nearest neighbor(KNN)and kernel support vector machine(KSVM)are applied as feature classifiers to verify the classification effect.Experimental results on the measured aircraft datasets show that KJDA can reduce the dimensionality,and improve target recognition performance.

DM-L Based Feature Extraction and Classifier Ensemble for Object Recognition

Deep Learning is a powerful technique that is widely applied to Image Recognition and Natural Language Processing tasks amongst many other tasks. In this work, we propose an efficient technique to utilize pre-trained Convolutional Neural Network (CNN) architectures to extract powerful features from images for object recognition purposes. We have built on the existing concept of extending the learning from pre-trained CNNs to new databases through activations by proposing to consider multiple deep layers. We have exploited the progressive learning that happens at the various intermediate layers of the CNNs to construct Deep Multi-Layer (DM-L) based Feature Extraction vectors to achieve excellent object recognition performance. Two popular pre-trained CNN architecture models i.e. the VGG_16 and VGG_19 have been used in this work to extract the feature sets from 3 deep fully connected multiple layers namely “fc6”, “fc7” and “fc8” from inside the models for object recognition purposes. Using the Principal Component Analysis (PCA) technique, the Dimensionality of the DM-L feature vectors has been reduced to form powerful feature vectors that have been fed to an external Classifier Ensemble for classification instead of the Softmax based classification layers of the two original pre-trained CNN models. The proposed DM-L technique has been applied to the Benchmark Caltech-101 object recognition database. Conventional wisdom may suggest that feature extractions based on the deepest layer i.e. “fc8” compared to “fc6” will result in the best recognition performance but our results have proved it otherwise for the two considered models. Our experiments have revealed that for the two models under consideration, the “fc6” based feature vectors have achieved the best recognition performance. State-of-the-Art recognition performances of 91.17% and 91.35% have been achieved by utilizing the “fc6” based feature vectors for the VGG_16 and VGG_19 models respectively. The recognition performance has been achieved by considering 30 sample images per class whereas the proposed system is capable of achieving improved performance by considering all sample images per class. Our research shows that for feature extraction based on CNNs, multiple layers should be considered and then the best layer can be selected that maximizes the recognition performance.

FEATURE EXTRACTION AND RECOGNITION FOR ECHOES OF HRR RADAR

This paper describes a novel target recognition scheme using High Range Resolution (HRR) radar signatures. AutoRegressive (AR) method is used to extract features from HRR radar echoes based on scattering center model of target. The optimal linear transformation based on Euclidian distribution distance criterion is performed on AR model parameter vectors to reduce dimension of feature vectors further and improve the class discrimination capability of feature vectors. The optimization algorithm is designed utilizing the quadratic property of criterion function and Gaussian kernel based Parzen window density function estimator. The concept of Stochastic Information Gradient (SIG) is incorporated into the gradient of cost function to decrease the computational complexity of the algorithm. Simulation results using three real airplanes,data show the effectiveness of the proposed method.

A SVM-Based Feature Extraction for Face Recognition

Social computing, a cross science of computational science and social science, is affecting people’s learning, work and life recently. Face recognition is going deep into every field of social life, and the feature extraction is particularly important. Linear Discriminant Analysis (LDA) is an effective feature extraction method. However, the traditional LDA cannot solve the nonlinear problem and small sample problem existing in high dimensional space. In this paper, the method of the Support Vector-based Direct Discriminant Analysis (SVDDA) is proposed. It incorporates SVM algorithm into LDA, extends SVM to nonlinear eigenspace, and optimizes eigenvalue to improve performance. Moreover, this paper combines SVDDA with the social computing theory. The experiments were tested on different face datasets. Compared with other existing methods, SVDDA has higher robustness and optimal performance.

A New Effective Method for Ship Target Recognition

In this paper, the problem of reliable automatic target recognition from incoherent radar returns is discussed and a new method for ship target recognition is proposed. Based on this method, an experimental system for ship target recognition is implemented. The results obtained from the theoretical and experimental study indicate that a high reliability of recognition can be achieved by using the designed recognition system. An average success rate of more than 90% is reached for 8 classes of ships.

A feature extraction method for synthetic aperture radar(SAR) automatic target recognition based on maximum interclass distance

Synthetic aperture radar(SAR) automatic target recognition is an important application in SAR.How to extract features has restricted the application of SAR technology seriously.In this paper,a new feature extraction method for SAR automatic target recognition based on maximum interclass distance is proposed,which integrates class and neighborhood information.This method can reinforce discriminative power using maximum interclass distance,so it can improve recognition rate effectively.

New supervised learning classifiers for structural damage diagnosis using time series features from a new feature extraction technique

The motivation for this article is to propose new damage classifiers based on a supervised learning problem for locating and quantifying damage.A new feature extraction approach using time series analysis is introduced to extract damage-sensitive features from auto-regressive models.This approach sets out to improve current feature extraction techniques in the context of time series modeling.The coefficients and residuals of the AR model obtained from the proposed approach are selected as the main features and are applied to the proposed supervised learning classifiers that are categorized as coefficient-based and residual-based classifiers.These classifiers compute the relative errors in the extracted features between the undamaged and damaged states.Eventually,the abilities of the proposed methods to localize and quantify single and multiple damage scenarios are verified by applying experimental data for a laboratory frame and a four-story steel structure.Comparative analyses are performed to validate the superiority of the proposed methods over some existing techniques.Results show that the proposed classifiers,with the aid of extracted features from the proposed feature extraction approach,are able to locate and quantify damage;however,the residual-based classifiers yield better results than the coefficient-based classifiers.Moreover,these methods are superior to some classical techniques.

The research on high speed underwater target recognition based on fuzzy logic inference

The underwater target recognition is a key technology in acoustic confrontation and underwater defence. In this article, a recognition system based on fuzzy logic inference (FLI) is set up. This system is mainly composed of three parts: the fuzzy input module, the fuzzy logic inference module with a set of inference rules and the de-fuzzy output module. The inference result shows the recognition system is effective in most conditions.

Ship recognition based on HRRP via multi-scale sparse preserving method

In order to extract the richer feature information of ship targets from sea clutter, and address the high dimensional data problem, a method termed as multi-scale fusion kernel sparse preserving projection(MSFKSPP) based on the maximum margin criterion(MMC) is proposed for recognizing the class of ship targets utilizing the high-resolution range profile(HRRP). Multi-scale fusion is introduced to capture the local and detailed information in small-scale features, and the global and contour information in large-scale features, offering help to extract the edge information from sea clutter and further improving the target recognition accuracy. The proposed method can maximally preserve the multi-scale fusion sparse of data and maximize the class separability in the reduced dimensionality by reproducing kernel Hilbert space. Experimental results on the measured radar data show that the proposed method can effectively extract the features of ship target from sea clutter, further reduce the feature dimensionality, and improve target recognition performance.

Principal Component Feature for ANN-Based Speech Recognition

利用函数逼近原理和主成份分析方法，提出了一种可用于解决语音信号时间规正和简化神经网络结构的语音信号主分量特征．该特征的提取过程模拟了人耳听觉系统的信息感知过程．

Application of Particle Swarm Optimization to Fault Condition Recognition Based on Kernel Principal Component Analysis

Panicle swarm optimization （PSO） is an optimization algorithm based on the swarm intelligent principle. In this paper the modified PSO is applied to a kernel principal component analysis （ KPCA ） for an optimal kernel function parameter. We first comprehensively considered within-class scatter and between-class scatter of the sample features. Then, the fitness function of an optimized kernel function parameter is constructed, and the particle swarm optimization algorithm with adaptive acceleration （CPSO） is applied to optimizing it. It is used for gearbox condi- tion recognition, and the result is compared with the recognized results based on principal component analysis （PCA）. The results show that KPCA optimized by CPSO can effectively recognize fault conditions of the gearbox by reducing bind set-up of the kernel function parameter, and its results of fault recognition outperform those of PCA. We draw the conclusion that KPCA based on CPSO has an advantage in nonlinear feature extraction of mechanical failure, and is helpful for fault condition recognition of complicated machines.

Radar Target Recognition Algorithm Based on RCS Observation Sequence——Set-Valued Identification Method

This paper studies the problem of radar target recognition based on radar cross section(RCS)observation sequence.First,the authors compute the discrete wavelet transform of RCS observation sequence and extract a valid statistical feature vector containing five components.These five components represent five different features of the radar target.Second,the authors establish a set-valued model to represent the relation between the feature vector and the authenticity of the radar target.By set-valued identification method,the authors can estimate the system parameter,based on which the recognition criteria is given.In order to illustrate the efficiency of the proposed recognition method,extensive simulations are given finally assuming that the true target is a cone frustum and the RCS of the false target is normally distributed.The results show that the set-valued identification method has a higher recognition rate than the traditional fuzzy classification method and evidential reasoning method.

基于YOLO v5的农田杂草识别轻量化方法研究

针对已有杂草识别模型对复杂农田环境下多种目标杂草的识别率低、模型内存占用量大、参数多、识别速度慢等问题,提出了基于YOLO v5的轻量化杂草识别方法。利用带色彩恢复的多尺度视网膜(Multi-scale retinex with color restoration, MSRCR)增强算法对部分图像数据进行预处理,提高边缘细节模糊的图像清晰度,降低图像中的阴影干扰。使用轻量级网络PP-LCNet重置了识别模型中的特征提取网络,减少模型参数量。采用Ghost卷积模块轻量化特征融合网络,进一步降低计算量。为了弥补轻量化造成的模型性能损耗,在特征融合网络末端添加基于标准化的注意力模块(Normalization-based attention module, NAM),增强模型对杂草和玉米幼苗的特征提取能力。此外,通过优化主干网络注意力机制的激活函数来提高模型的非线性拟合能力。在自建数据集上进行实验,实验结果显示,与当前主流目标检测算法YOLO v5s以及成熟的轻量化目标检测算法MobileNet v3-YOLO v5s、ShuffleNet v2-YOLO v5s比较,轻量化后杂草识别模型内存占用量为6.23 MB,分别缩小54.5%、12%和18%;平均精度均值(Mean average precision, mAP)为97.8%,分别提高1.3、5.1、4.4个百分点。单幅图像检测时间为118.1 ms,达到了轻量化要求。在保持较高模型识别精度的同时大幅降低了模型复杂度,可为采用资源有限的移动端设备进行农田杂草识别提供技术支持。

融合人体骨架和姿势信息特征的轻量级人体动作识别方法

针对人体动作识别任务中特征值选取不当导致识别率低、使用多模态数据导致训练成本高等问题,提出一种轻量级人体动作识别方法。首先使用OpenPose、PoseNet提取出人体骨架信息,使用BWT69CL传感器提取姿势信息;其次对数据进行预处理、特征融合,对人体动作进行深度学习分类识别;最后,为验证此方法的有效性,在公开数据集WISDM、UCIHAR、HASC和自建的人体动作数据集上进行实验验证,并使用改进的目标引导注意力机制(target-guided attention,TGA)–长短期记忆(long short term memory,LSTM)网络输出最终的分类结果。实验结果表明,在自建数据集下融合姿势和骨架特征达到99.87%准确率,相比于只使用姿势信息特征,识别准确率提高了约5.31个百分点;相比于只使用人体骨架特征,识别准确率提高了约1.87个百分点;在识别时间上相比于只使用姿势信息,识别时间降低了约29.73 s;相比于只使用人体骨架数据,识别时间降低了约9 s。使用该方法能及时有效地反映人体的运动意图,有助于提高人体动作和行为的识别准确率和训练效率。

基于面积加权GWT-GFT的水声目标识别

由于海洋环境的复杂性,水声目标的识别具有很大的挑战性。为解决这类复杂环境下特征提取的问题,提出了一种基于面积加权的图小波变换-图傅里叶变换(GWT-GFT)的分析方法。在完成数据预处理后,为了能够凸显顶点之间的关系,提出了一种新的基于顶点三角形面积的加权方法来构建图信号;构建好的图信号通过GWT分解为多尺度图分量;然后,利用GFT将这些分量从图域变换到特征值谱域进行分析;在此基础上,提取各分量特征值谱的特征;最后,利用基于高斯核函数的支持向量机(SVM)对获取的特征向量进行分类。基于水声信号ShipsEar数据库,采用5折交叉验证方法进行验证。与现有的其它方法相比,所提的模型以36个特征在376656个样本上取得了97.22%的准确率,证明了该分析方法的有效性和鲁棒性。

复杂环境下的轻量化道路目标识别算法研究

道路目标识别是智能交通系统解决城市拥堵问题的核心技术之一,然而现有算法在复杂交通环境下识别效果较差,存在大量漏检和误检情况,且模型参数量大,不适合在实际场景下部署于资源有限的移动端设备。针对以上问题,提出一种复杂环境下的轻量化道路目标识别算法。基于SSD算法结构设计一种可重构的特征提取网络框架,利用3种轻量化模块分别构建浅层特征提取网络,以自定义的Additional Block构建深层特征提取网络,并分别采用通道注意力机制和轻量化感受野扩大(RFB-L)模块提升模型对各尺寸目标的检测效果。利用自定义的像素与通道信息融合模块实现浅层与深层特征的融合,丰富待检测特征图包含的信息。同时,提出一种多特征融合的学习率调节算法,使得训练过程中模型性能稳定地达到收敛。自制复杂拥堵道路数据集Hohhot_city用于算法训练和测试,与主流算法的对比实验结果表明,该算法性能明显优于参数量同级别的YOLOv4-tiny和YOLOv5s算法,在参数量不到YOLOv5m算法40%的情况下与其检测精度接近,并取得了12.8 ms的推理时间和99.1%的均值平均精度。