目的红外小目标检测旨在从复杂背景中准确识别和定位弱小红外目标,在海上侦查、军事救援等任务中具有重要的应用前景。然而,受限于红外图像中目标尺寸小、对比度低等因素,当前的检测方法仍难以实现检测精度与误报率之间的平衡。针对上...目的红外小目标检测旨在从复杂背景中准确识别和定位弱小红外目标,在海上侦查、军事救援等任务中具有重要的应用前景。然而,受限于红外图像中目标尺寸小、对比度低等因素,当前的检测方法仍难以实现检测精度与误报率之间的平衡。针对上述问题,提出一种基于选择性注意力的红外小目标检测网络(selective attentionbased network for infrared small target detection,SANet)。方法具体而言,设计了双路径语义感知模块,用于增强网络对弱小目标的感知能力。该模块融合了标准卷积与风车型卷积两种路径,兼顾局部空间一致性与全局上下文感知能力,并进一步引入空间/通道注意力机制以细化特征表达,从而有效提升了目标与背景的可区分性。此外,为克服U-Net中静态跳跃连接在特征融合中的局限性,进一步提出了选择性注意力融合模块。该模块基于空间动态权重机制实现跨尺度特征的自适应融合,能够根据空间显著性选择性增强关键区域,从而提升对真实目标与误报的辨识能力。结果在3个公开基准数据集上的实验结果验证了所提出的SANet在交并比(intersection over union,IoU)、nIoU、P_(d)和F_(a)4个指标上均优于现有先进方法,其中,本文方法的IoU指标在上述基准数据集上比次优方法分别提升1.93%、4.32%和2.21%。结论SANet有效增强了网络对小目标的感知能力、关键特征的表达能力以及背景干扰的抑制能力。源代码可以在https://gitcode.com/m0_61988291/SANet上获取。展开更多
This paper is concerned with the robust control synthesis of autonomous underwater vehicle （AUV） for general path following maneuvers. First, we present maneuvering kinematics and vehicle dynamics in a unified frame...This paper is concerned with the robust control synthesis of autonomous underwater vehicle （AUV） for general path following maneuvers. First, we present maneuvering kinematics and vehicle dynamics in a unified framework. Based on H∞ loop-shaping procedure, the 2-DOF autopilot controller has been presented to enhance stability and path tracking. By use of model reduction, the high-order control system is reduced to one with reasonable order, and further the scaled low-order controller has been analyzed in both the frequency and the time domains. Finally, it is shown that the autopilot control system provides robust performance and stability against prescribed levels of uncertainty.展开更多
A series of dynamic behavior tests on Nanjing flake-shaped fine sand were performed by using the WFI cyclic triaxial apparatus made in England. The dynamic behaviors of Nanjing flake-shaped fine sand under different s...A series of dynamic behavior tests on Nanjing flake-shaped fine sand were performed by using the WFI cyclic triaxial apparatus made in England. The dynamic behaviors of Nanjing flake-shaped fine sand under different static deviator stress levels and cyclic stress ratios were studied. Through comparing the effective stress path under cyclic loading with static loading, the processes of liquefaction of saturated Nanjing flake-shaped fine sand with development of dynamic pore-water pressure, including the initial compact state, compression state and dilative state, were investigated. The variation of the shear stiffness with the number of cycles and cyclic strain was investigated by analyzing the secant shear modulus in each unload-reload loop of dynamic stress-strain relationship. And by means of the exponential function, the empirical equations of the relationship between secant shear modulus Gsec, shear modulus ratio Gsec/Gmax and cyclic strain ε were established based on series of test results. The results show that according to different combinations of static deviator stress and cyclic stress, two kinds of failure patterns with deviator stress reversal or no deviator stress reversal are observed in the samples tested in this series, including cyclic mobility and the failure of accumulation residual strain. In addition, the degradation of dynamic shear modulus is due to the development of vibration pore-water pressure and it is observed that the shear modulus reduces with the progressive number of cycles.展开更多
文摘目的红外小目标检测旨在从复杂背景中准确识别和定位弱小红外目标,在海上侦查、军事救援等任务中具有重要的应用前景。然而,受限于红外图像中目标尺寸小、对比度低等因素,当前的检测方法仍难以实现检测精度与误报率之间的平衡。针对上述问题,提出一种基于选择性注意力的红外小目标检测网络(selective attentionbased network for infrared small target detection,SANet)。方法具体而言,设计了双路径语义感知模块,用于增强网络对弱小目标的感知能力。该模块融合了标准卷积与风车型卷积两种路径,兼顾局部空间一致性与全局上下文感知能力,并进一步引入空间/通道注意力机制以细化特征表达,从而有效提升了目标与背景的可区分性。此外,为克服U-Net中静态跳跃连接在特征融合中的局限性,进一步提出了选择性注意力融合模块。该模块基于空间动态权重机制实现跨尺度特征的自适应融合,能够根据空间显著性选择性增强关键区域,从而提升对真实目标与误报的辨识能力。结果在3个公开基准数据集上的实验结果验证了所提出的SANet在交并比(intersection over union,IoU)、nIoU、P_(d)和F_(a)4个指标上均优于现有先进方法,其中,本文方法的IoU指标在上述基准数据集上比次优方法分别提升1.93%、4.32%和2.21%。结论SANet有效增强了网络对小目标的感知能力、关键特征的表达能力以及背景干扰的抑制能力。源代码可以在https://gitcode.com/m0_61988291/SANet上获取。
基金a part of the project titled "Development of Key Marine Equipments for Enhancement of Ocean Industry-Development of Underwater Manipulator and Thrusting System Driven by Electric Motor" funded by the Ministry of Land, Transport and Maritime Affairs, Korea
文摘This paper is concerned with the robust control synthesis of autonomous underwater vehicle （AUV） for general path following maneuvers. First, we present maneuvering kinematics and vehicle dynamics in a unified framework. Based on H∞ loop-shaping procedure, the 2-DOF autopilot controller has been presented to enhance stability and path tracking. By use of model reduction, the high-order control system is reduced to one with reasonable order, and further the scaled low-order controller has been analyzed in both the frequency and the time domains. Finally, it is shown that the autopilot control system provides robust performance and stability against prescribed levels of uncertainty.
基金Project (2007CB714200) supported by National Basic Research Program of ChinaProject (90715018) supported by Key Research Project of National Natural Science Foundation of China
文摘A series of dynamic behavior tests on Nanjing flake-shaped fine sand were performed by using the WFI cyclic triaxial apparatus made in England. The dynamic behaviors of Nanjing flake-shaped fine sand under different static deviator stress levels and cyclic stress ratios were studied. Through comparing the effective stress path under cyclic loading with static loading, the processes of liquefaction of saturated Nanjing flake-shaped fine sand with development of dynamic pore-water pressure, including the initial compact state, compression state and dilative state, were investigated. The variation of the shear stiffness with the number of cycles and cyclic strain was investigated by analyzing the secant shear modulus in each unload-reload loop of dynamic stress-strain relationship. And by means of the exponential function, the empirical equations of the relationship between secant shear modulus Gsec, shear modulus ratio Gsec/Gmax and cyclic strain ε were established based on series of test results. The results show that according to different combinations of static deviator stress and cyclic stress, two kinds of failure patterns with deviator stress reversal or no deviator stress reversal are observed in the samples tested in this series, including cyclic mobility and the failure of accumulation residual strain. In addition, the degradation of dynamic shear modulus is due to the development of vibration pore-water pressure and it is observed that the shear modulus reduces with the progressive number of cycles.
