夜间成像光测数字图像增强技术研究

分析夜间成像的光测数字图像灰度范围、对比度和信息量等主要特点,为图像增强的策略选择和效果评价提供依据。针对夜间成像光测数字图像灰度范围较窄等特点,提出局部灰度修正的增强方法适当增强图像对比度,又避免细节损失,同时,利用拉普拉斯锐化突出目标轮廓和边缘,并利用人眼对彩色敏感的特点,采用伪彩色增强提高对图像中模糊目标的分辨能力。通过上述3种增强技术,提高了对夜间成像暗图像的目标识别能力,有利于后续判读处理。     

一种低分辨雷达飞机目标识别方法

介绍了一种基于低分辨雷达回波序列的飞机目标识别方法。该方法把一维回波序列变换成二维轮廓像,在二维轮廓像基础上,提取到较一维回波更加稳定、有效的特征。采用神经网络做为分类器,利用所提取的特征对五类飞机目标进行识别试验,得到了较高的正确识别率。     

图像信号预处理的一种算法

提出了针对含有斑、点目标的遥感图像进行有损压缩的一种方法,介绍了去局部均值算法原理,依据3σ准则推导了门限取值公式,给出了算法的流程。通过仿真实验,分析了算法的有效性。     

基于最小轨道交叉距离的小行星顺访探测目标预筛选方法

小行星探测有助于研究太阳系演化等重要科学问题,在深空探测任务转移途中实施小行星顺访探测可增加科学研究回报。直接通过轨道递推筛选小行星探测目标计算量大、效率低,针对该问题提出了基于最小轨道交叉距离的目标预筛选方法。在推导出适用于计算双曲线轨道的最小轨道交叉距离公式后,将此理论应用到小行星顺访探测目标筛选中。首先基于探测器与小行星轨道的形状、空间位置计算二者轨道在空间中的几何最近距离,预筛选出可能满足接近距离标准的小行星目标;然后基于轨道递推模型,筛选出真实最近距离小于可接近标准的目标小行星。仿真结果显示,基于最小轨道交叉距离的预筛选方法可有效减少计算量,降低计算时间,提高小行星顺访目标筛选的效率。     

Large-scale real-world radio signal recognition with deep learning

In the past ten years, many high-quality datasets have been released to support the rapid development of deep learning in the fields of computer vision, voice, and natural language processing. Nowadays, deep learning has become a key research component of the Sixth-Generation wireless systems (6G) with numerous regulatory and defense applications. In order to facilitate the application of deep learning in radio signal recognition, in this work, a large-scale real-world radio signal dataset is created based on a special aeronautical monitoring system - Automatic Dependent Surveillance-Broadcast (ADS-B). This paper makes two main contributions. First, an automatic data collection and labeling system is designed to capture over-the-air ADS-B signals in the open and real-world scenario without human participation. Through data cleaning and sorting, a high-quality dataset of ADS-B signals is created for radio signal recognition. Second, we conduct an in-depth study on the performance of deep learning models using the new dataset, as well as comparison with a recognition benchmark using machine learning and deep learning methods. Finally, we conclude this paper with a discussion of open problems in this area.     

KDTGAN：基于Transformer-GAN和知识蒸馏的高光谱目标检测

高光谱目标检测在地球观测中至关重要,被广泛应用于军事和民用领域。然而,由于高光谱图像的背景复杂性和目标样本的有限性,该任务面临较大的挑战。本文首先采用CEM（约束能量最小化）粗检测方法提取背景数据。随之,引入了一种新的知识蒸馏模型,即KDTGAN（通过Transformer-GAN实现）。教师模型的生成器采用了Transformer编码器的结构,并结合多尺度数据融合的方法,能够准确地学习背景分布,进而通过重构背景信息实现目标检测。为了克服GAN（生成对抗网络）训练不稳定的挑战,特别是纯背景数据的稀缺性,本文提出了一种新的损失算法,以减小可疑目标样本对模型性能的负面影响。为了降低模型的计算负担,本文引入知识蒸馏,并设计新的蒸馏损失对学生模型加以约束,使模型轻量化的同时提高学生模型检测精度。实验结果表明：KDTGAN相较于当前检测方法表现更优,具有更高的检测精度和鲁棒性。     

Moving scanning emitter tracking by a single observer using time of interception: Observability analysis and algorithm

The target motion analysis (TMA) for a moving scanning emitter with known fixed scan rate by a single observer using the time of interception (TOI) measurements only is investigated in this paper.By transforming the TOI of multiple scan cycles into the direction difference of arrival (DDOA) model,the observability analysis for the TMA problem is performed.Some necessary conditions for uniquely identifying the scanning emitter trajectory are obtained.This paper also proposes a weighted instrumental variable (WIV) estimator for the scanning emitter TMA,which does not require any initial solution guess and is closed-form and computationally attractive.More importantly,simulations show that the proposed algorithm can provide estimation mean square error close to the Cramer-Rao lower bound (CRLB) at moderate noise levels with significantly lower estimation bias than the conventional pseudo-linear least square (PLS) estimator.     

Target-enclosing affine formation control of two-layer networked spacecraft with collision avoidance

This paper addresses a target-enclosing problem for multiple spacecraft systems by proposing a two-layer affine formation control strategy. Compared with the existing methods, the adopted two-layer network structure in this paper is generally directed, which is suitable for practical space missions. Firstly, distributed finite-time sliding-mode estimators and formation controllers in both layers are designed separately to improve the flexibility of the formation control system. By introducing the properties of affine transformation into formation control protocol design, the controllers can be used to track different time-varying target formation patterns. Besides, multi-layer time-varying encirclements can be achieved with particular shapes to surround the moving target. In the sequel, by integrating adaptive neural networks and specialized artificial potential functions into backstepping controllers, the problems of uncertain Euler-Lagrange models, collision avoidance as well as formation reconfiguration are solved simultaneously. The stability of the proposed controllers is verified by the Lyapunov direct method. Finally, two simulation examples of triangle formation and more complex hexagon formation are presented to illustrate the feasibility of the theoretical results.     

复杂环境下一种基于改进核相关滤波的视觉鲁棒目标跟踪方法

目标跟踪在自动驾驶和智能监控系统等实时视觉应用中发挥着重要作用。在遮挡、相似干扰等情况下，传统的基于相关滤波的跟踪算法容易发生漂移，鲁棒性有待进一步提高。基于此，提出了一种扩展特征描述的检测辅助核相关滤波目标跟踪架构。首先，在传统的核相关滤波目标跟踪算法的基础上，通过目标检测辅助对跟踪结果进行质量判断，实现对遮挡以及目标丢失的判别；然后通过拓展特征模板的构建与匹配，实现抗干扰相似目标判断及目标重定位；最终，以行人跟踪为例进行了试验，分别通过OTB数据及验证试验和移动机器人平台视觉跟踪验证试验，验证了算法的可行性，并对算法的跟踪性能进行了测试。试验结果表明，所提方法能够稳定地跟踪移动目标，对遮挡、相似干扰具有较强的鲁棒性。     

含无序量测的多传感器目标跟踪滤波算法

由于通信链路的随机时间延迟和星上传感器测量的预处理时间的不同等因素,导致在目标飞行器的测量中产生无序量测现象。为解决该问题,提出了一种基于扩展卡尔曼滤波的前向预测多步滞后无序量测处理算法。该算法首先采用扩展卡尔曼滤波算法估算出目标飞行器的状态方程和协方差方程,然后在滤波过程中利用前向滤波更新的方法,将协方差方程更新结果去相关后,累积到当前协方差方程滤波结果中,从而有效解决了目标飞行器测量中的无序量测问题。最后,将该算法与扩展卡尔曼滤波算法、丢弃算法进行了对比仿真。仿真结果表明,采用该算法处理目标飞行器的位置和飞行速度,得到的测量误差较小,在整个观测时间内,测量误差的收敛性较好,能够实现对目标飞行器的精确测量和跟踪。