Fast-armored target detection based on multi-scale representation and guided anchor

Focused on the task of fast and accurate armored target detection in ground battlefield,a detection method based on multi-scale representation network(MS-RN) and shape-fixed Guided Anchor(SF-GA)scheme is proposed.Firstly,considering the large-scale variation and camouflage of armored target,a new MS-RN integrating contextual information in battlefield environment is designed.The MS-RN extracts deep features from templates with different scales and strengthens the detection ability of small targets.Armored targets of different sizes are detected on different representation features.Secondly,aiming at the accuracy and real-time detection requirements,improved shape-fixed Guided Anchor is used on feature maps of different scales to recommend regions of interests(ROIs).Different from sliding or random anchor,the SF-GA can filter out 80% of the regions while still improving the recall.A special detection dataset for armored target,named Armored Target Dataset(ARTD),is constructed,based on which the comparable experiments with state-of-art detection methods are conducted.Experimental results show that the proposed method achieves outstanding performance in detection accuracy and efficiency,especially when small armored targets are involved.

High-Precision Vital Signs Detection Method Based on Spectrum Refinement and Extended DCMA

In this paper,the spectral estimation algorithm is extended to the detection of human vi-tal signs by mm-wave frequency modulated continuous wave(FMCW)radar,and a comprehensive algorithm based on spectrum refinement and the extended differentiate and cross multiply al-gorithm(DCMA)has been proposed.Firstly,the improved DFT algorithm is used to accurately obtain the distance window of human body.Secondly,phase ambiguity in phase extraction is avoided based on extended DCMA algorithm.Then,the spectrum range of refinement is determ-ined according to the peak position of the spectrum,and the respiratory and heartbeat frequency information is obtained by using chirp z-transform(CZT)algorithm to perform local spectrum re-finement.For verification,this paper has simulated the radar echo signal modulated by the simu-lated cardiopulmonary signal according to the proposed algorithm.By recovering the simulated car-diopulmonary signal,the high-precision respiratory and heartbeat frequency have been obtained.The results show that the proposed algorithm can effectively restore human breathing and heart-beat signals,and the relative error of frequency estimation is basically kept below 1.5%.

Flaw Detection Capability and Sensitivity in the Inspection of Nuclear Containment Liner and Shell Mock-Ups Utilizing a Magnetostrictive Sensor (MsS) Guided Wave UT Technique

This paper describes the results of a project on the inspection of visually inaccessible areas of nuclear containment liners and shells via the advanced Magnetostrictive sensor (MsS) Guided Wave (GW) nondestructive inspection technique. Full scale mockups that simulated shell and liner regions of interest in the containment of both a Pressurized Water Reactor (PWR) and Boiling Water Reactor (BWR) were constructed. Inspections were performed on the mock-ups in three stages to discern the signal attenuation caused by flaws and caused by concrete in the structures. The effect of concrete being in close proximity to the liner and shell was determined, and the capability to detect and size flaws via this GW technique was evaluated.

The prediction formula and a risk-based sensor scheduling method in target detection with guiding information

Sensor scheduling is used to improve the sensing performance in the estimation of targets’states.However,few papers are on the sensor scheduling for target detection with guiding information.This letter can remedy this deficiency.A risk-based target detection method with guiding information is provided firstly,based on which,the sensor scheduling approach is aiming at reducing the risk and uncertainty in target detection,namely risk-based sensor scheduling method.What should be stressed is that the Prediction Formula in sensor scheduling is proposed.Lastly,some examples are conducted to stress the effectiveness of this proposed method.

DETECTION OF LONGITUDINAL DEFECT IN PIPES USING TORSIONAL MODES

The multi-modes and disperse characteristics of torsional modes in pipes are investigated theoretically and experimentally. At all frequencies, both phase velocity and group velocity of the lowest torsional mode T（0,1） are constant and equal to shear wave velocity. T（0,1） mode at all frequencies is the fastest torsional mode. In the experiments, T（0,1） mode is excited and received in pipes using 9 thickness shear vibration mode piezoelectric ceramic elements. Furthermore, an artificial longitudinal defect of a 4 m long pipe is detected using T（0,1） mode at 50 kHz. Experimental results show that it is feasible for longitudinal defect detection in pipes using T（0,1） mode of ultrasonic guided waves.

An improved bicubic imaging fitting algorithm for 3D radar detection target

3D ground-penetrating radar has been widely used in urban road underground disease detection due to its nondestructive,efficient,and intuitive results.However,the 3D imaging of the underground target body presents the edge plate phenomenon due to the space between the 3D radar array antennas.Consequently,direct 3D imaging using detection results cannot reflect underground spatial distribution characteristics.Due to the wide-beam polarization of the ground-penetrating radar antenna,the emission of electromagnetic waves with a specific width decreases the strong middle energy on both sides gradually.Therefore,a bicubic high-precision 3D target body slice-imaging fitting algorithm with changing trend characteristics is constructed by combining the subsurface target characteristics with the changing spatial morphology trends.Using the wide-angle polarization antenna’s characteristics in the algorithm to build the trend factor between the measurement lines,the target body change trend and the edge detail portrayal achieve a 3D ground-penetrating radar-detection target high-precision fitting.Compared with other traditional fitting techniques,the fitting error is small.This paper conducts experiments and analyses on GpaMax 3D forward modeling and 3D ground-penetrating measured radar data.The experiments show that the improved bicubic fitting algorithm can eff ectively improve the accuracy of underground target slice imaging and the 3D ground-penetrating radar’s anomaly interpretation.

Guided Dropout: Improving Deep Networks Without Increased Computation

Deep convolution neural networks are going deeper and deeper.How-ever,the complexity of models is prone to overfitting in training.Dropout,one of the crucial tricks,prevents units from co-adapting too much by randomly drop-ping neurons during training.It effectively improves the performance of deep net-works but ignores the importance of the differences between neurons.To optimize this issue,this paper presents a new dropout method called guided dropout,which selects the neurons to switch off according to the differences between the convo-lution kernel and preserves the informative neurons.It uses an unsupervised clus-tering algorithm to cluster similar neurons in each hidden layer,and dropout uses a certain probability within each cluster.Thereby this would preserve the hidden layer neurons with different roles while maintaining the model’s scarcity and gen-eralization,which effectively improves the role of the hidden layer neurons in learning the features.We evaluated our approach compared with two standard dropout networks on three well-established public object detection datasets.Experimental results on multiple datasets show that the method proposed in this paper has been improved on false positives,precision-recall curve and average precision without increasing the amount of computation.It can be seen that the increased performance of guided dropout is thanks to shallow learning in the net-works.The concept of guided dropout would be beneficial to the other vision tasks.

Robust Damage Detection and Localization Under Complex Environmental Conditions Using Singular Value Decomposition-based Feature Extraction and One-dimensional Convolutional Neural Network

Ultrasonic guided wave is an attractive monitoring technique for large-scale structures but is vulnerable to changes in environmental and operational conditions(EOC),which are inevitable in the normal inspection of civil and mechanical structures.This paper thus presents a robust guided wave-based method for damage detection and localization under complex environmental conditions by singular value decomposition-based feature extraction and one-dimensional convolutional neural network(1D-CNN).After singular value decomposition-based feature extraction processing,a temporal robust damage index(TRDI)is extracted,and the effect of EOCs is well removed.Hence,even for the signals with a very large temperature-varying range and low signal-to-noise ratios(SNRs),the final damage detection and localization accuracy retain perfect 100%.Verifications are conducted on two different experimental datasets.The first dataset consists of guided wave signals collected from a thin aluminum plate with artificial noises,and the second is a publicly available experimental dataset of guided wave signals acquired on a composite plate with a temperature ranging from 20℃to 60℃.It is demonstrated that the proposed method can detect and localize the damage accurately and rapidly,showing great potential for application in complex and unknown EOC.

A New High-Precision and Lightweight Detection Model for Illegal Construction Objects Based on Deep Learning

Illegal construction has caused serious harm around the world. However, current methods are difficult to detect illegal construction activities in time, and the calculation complexity and the parameters of them are large. To solve these challenges, a new and unique detection method is proposed, which detects objects related to illegal buildings in time to discover illegal construction activities. Meanwhile, a new dataset and a high-precision and lightweight detector are proposed. The proposed detector is based on the algorithm You Only Look Once (YOLOv4). The use of DenseNet as the backbone of YDHNet enables better feature transfer and reuse, improves detection accuracy, and reduces computational costs. Meanwhile, depthwise separable convolution is employed to lightweight the neck and head to further reduce computational costs. Furthermore, H-swish is utilized to enhance non-linear feature extraction and improve detection accuracy. Experimental results illustrate that YDHNet realizes a mean average precision of 89.60% on the proposed dataset, which is 3.78% higher than YOLOv4. The computational cost and parameter count of YDHNet are 26.22 GFLOPs and 16.18 MB, respectively. Compared to YOLOv4 and other detectors, YDHNet not only has lower computational costs and higher detection accuracy, but also timely identifies illegal construction objects and automatically detects illegal construction activities.

Detection of long bones based on Barker code excited ultrasonic guided waves

Single pulse excited ultrasonic guided wave surfers high attenuation during the propagation in long bones.This results in small amplitude and low signal-to-noise ratio（SNR）of measured signals.Thus,the Barker code excitation is introduced into long bone detection to improve the quality of received signals,due to its efficiency in increasing amplitude and SNR.Both simulation and in vitro experiment were performed,and the results were decoded by the weighted match filter（WMF） and the finite impulse response- least squares inverse filter（FIRLSIF）,respectively.The comparison between the results of Barker code excitation and sine pulse excitation was presented.For 13-bit Barker code excitation,WMF produced 13 times larger amplitude than sine pulse excitation,while FIR-LSIF achieved higher peak-sidelobe-level（PSL） of -63.59 dB and better performance in noise suppression.The results show that the Barker code excited guided waves have the potential to be applied to the long bone detection.

全自动检测系统在医用硬式内窥镜检测中的应用研究

目的评价使用全自动硬式内窥镜检测系统对硬式内窥镜光学性能检测的优势,并分析影响硬式内窥镜功能减退的潜在因素。方法选取某三级甲等综合医院2023年5—7月处理的硬式内窥镜镜头78条、导光束161条为研究对象,先后采用目测法和仪器检测法对其进行检测。仪器检测指标包括视向角、视场角、照度、景深、导光率、对焦、内窥镜内部密封性及纯净度;导光束指标包括光纤照度、导光率及色彩测试,比较2种检测方法的结果差异。结果目测法与仪器检测法检测硬式内窥镜镜头的合格率分别为94.87%和73.08%,差异有统计学意义(P<0.05);导光束的合格率分别为98.75%和96.89%,差异无统计学意义(P>0.05);仪器检测结果显示硬式内窥镜镜头的功能障碍主要集中在导光率和照度两方面。结论针对内窥镜镜头检测,仪器检测法明显优于目测法;针对导光束检测,两种检测方法未有显著差异。硬式内窥镜检测系统有助于及早发现硬式内窥镜镜头内部的结构变化和功能减退情况,是一种有助于降低内镜侵入诊疗操作风险、节约医院运营成本的科学检测工具。

低频导波检测技术在输油场站应用的效果分析

本文主要是对低频导波检测技术原理和应用进行分析,进而选取本次最优开挖验证点,通过MsS长距离超声导波检测仪在姬塬外输总站管网腐蚀检测过程中的应用效果分析,判断其应用效果具有哪些局限性和优缺点,为后续输油场站常规检测工作提供参考依据。

基于信号响应分析模型的金属结构损伤导波检出概率

飞机结构损伤导波在线监测技术作为一种新颖的无损检测手段,为了真正实现该技术在结构运营维护过程中的视情维护,必须明确其结构损伤检出概率(probability of detection,POD),以指导结构检查维修方案的制定。提出了一种基于信号响应分析模型的结构损伤导波POD计算方法,该方法通过构建在线导波监测信号的损伤指数与裂纹长度间的对应关系,得到结构损伤POD的统计计算模型,并分析了拟合参数的不确定性对计算模型的影响,构建了不同置信度下的导波POD计算模型。通过开展金属开孔和搭接结构疲劳裂纹导波监测试验,验证了该方法的有效性。试验结果表明,损伤指数类型、对应关系拟合函数和传感器监测方案均对结构损伤导波POD具有影响,且在95%置信度90%POD下金属开孔和搭接结构的可检裂纹长度分别约为2.6 mm和9.5 mm。

基于双分支特征聚合网络的车辆检测算法

车辆目标检测是自动驾驶的重要环节,现有的车辆目标检测算法在特征提取方面没有充分考虑卷积神经网络(convolutional neural network,CNN)和Transformer各自的优缺点,一定程度上限制了网络的整体性能。提出了一种由CNN和Transformer组成的双分支特征聚合网络。在编码阶段,基于CNN和Transformer各自的优势,构建了双分支主干网络来提取原始图像的特征信息;通过设计的多级别空间注意力模块和双支路特征聚合模块,使两个分支间的特征信息相互引导学习;通过构建的双分支注意力模块来进一步减少深层神经网络中特征信息的丢失。在实验部分通过消融实验和对比实验进一步验证了所提算法的有效性,其相比主流的目标检测算法,在mAP(mean average precision)指标上提升了约3.5%。

多种超声影像技术在乳腺癌诊断中的研究进展

乳腺癌是我国女性发病率最高的恶性肿瘤,常规超声是乳腺癌筛查常用的影像检查方法。弹性成像、超声造影、自动乳腺超声成像系统、S-detect技术及超声引导下穿刺活检等新技术正在飞速发展并逐渐应用于临床,为乳腺癌的诊断提供了更有价值的信息。本文就多种超声影像技术诊断乳腺癌的临床应用进展进行综述。

基于多模态交互引导网络的遥感图像密集车辆检测

遥感图像目标检测是大规模拥挤城市场景的风险评估和救援的技术基础,面临着光照干扰、遮挡、密集、拥挤等挑战。现有工作主要基于可见光单模态的遥感数据,对密集目标的特征表达及可分离能力受限。提出了一种面向遥感车辆检测的多模态交互引导网络,通过构建双流网络及多模态交互引导学习机制,显著提升密集车辆目标的可分离度,解决由于类间距离小、类内距离大,导致密集车辆的检测性能差的难题。在Potsdam公开数据集和DLR音乐节数据上的实验表明了该算法的鲁棒性和有效性。

联合光谱和空间双尺度细节注入的遥感图像融合

设计了一种基于引导滤波增强的联合光谱尺度和空间尺度细节注入的自适应遥感图像融合方法。首先,将全色图像作为引导图像,利用引导滤波对多光谱图像分区域处理,锐化纹理丰富区域、平滑光谱丰富区域,得到增强多光谱图像。其次,利用经典成分替换法和多尺度分析法分别提取光谱尺度和空间差异的细节信息,通过互信息整合得到包含双尺度的细节图像。再次,以原始多光谱图像和亮度分量的光谱相关关系、增强光谱图像的边缘信息作为细节图像的注入约束条件,得到细节注入图像。最后,将多光谱图像与联合空间和光谱双尺度的细节注入图像相加得到融合的高分辨率多光谱图像。分别利用IKONOS、QuickBird、WorldView4和高分二号四类遥感数据进行实验,结果表明,与其他融合方法相比,所提方法在主观视觉效果和客观量化指标上均表现出良好的性能,验证了所提方法的有效性。

基于FPGA的数字带通滤波器设计与仿真

为快速准确地提取超声导波信号中的目标信息,采用流水线加法树阵列结构设计低阶滤波器,采用直接型低阶叠加实现20阶高精度带通滤波器。运用流水线技术与FPGA片上资源相结合实现对该结构数字算法优化,并将其应用在镍铬合金超声导波测温系统中。结果表明:所设计的滤波器对512点的混频超声导波信号滤波耗时约1047个时钟。该滤波器总体设计简单,可以达到实时滤波的功能。

压电超声SH导波管道测厚方法优化及系统设计

为深入探究水平剪切(SH)波在有界平板中的传播特性,该研究从SH波的基本理论出发,根据无限大平板中传播的SH0波的非频散特性,开展有界平板结构中SH0波传播特性的仿真分析和实验验证。结果表明,在频厚积参数为3.0 MHzmm的激励条件下,有界平板结构中传播的SH0波能够保持优良的非频散特性。在此研究基础上,针对利用SH导波在测厚研究中存在的耦合效果不佳、回波信号杂乱等共性问题,借助平板型波导条设计了一套基于压电超声SH0导波的管道壁厚测量装置,并基于该测厚设备开展了管道壁厚检测实验。实验结果表明,所设计的“预固定”加“二次固定”耦合方法有效解决了耦合难题;基于该测厚装置提出了忽略首次回波信号的管道壁厚计算方法,该方法能够从原始回波信号中更加精确地计算出壁厚值,将测量值精度误差控制在0.05 mm以内。该文提出的研究方法为工业管道的壁厚检测和长期在线监测提供了新的解决思路,在工程实践中具有重要的参考价值和指导意义。

航天器结构中导波健康监测技术的若干进展

基于航天器结构健康监测需求,从空间碎片撞击定位、结构损伤评估、泄漏定位3个方面,阐述了导波的相关应用进展。首先分析了空间站、可重复使用运载器和燃料贮箱等结构面临的风险和监测需求,然后介绍了相关的理论和监测方法研究进展,以及多物理场导波建模和传播仿真技术,最后探讨了设计集成“撞、漏、损”多功能一体化结构健康监测系统的实现思路,以及导波结构健康监测技术在与空间机器人、空间激光溯源、无线通信和能量传输结合方面的应用前景。