基于YOLO模型的SAR舰船目标检测方法研究

随着海上交通运输业业务需求的不断增加,传统的目标检测方法已无法满足实际需求。由于卫星遥感技术的快速发展,基于合成孔径雷达（Synthetic Aperture Radar,SAR）图像舰船目标自动识别具有显著的应用潜力。近年来,深度学习技术在目标检测领域逐渐显现出优势,特别是YOLO (You Only Look Once)模型以其较高的精度和计算效率,为SAR舰船目标的识别提供了一种新的方法。为对比不同的YOLO模型在舰船目标识别领域的性能及其相比于两阶段深度学习算法的优势,本文首先对YOLO系列的结构进行了归纳总结;其次对当前广泛使用的数据集进行了对比分析,并基于SAR图像数据集（SAR Ship Detection Dataset,SSDD）的样本进行重新标注构建出本文的数据集;然后将YOLO系列模型与两阶段目标检测方法——更快速的区域卷积神经网络（Faster Region-based Convolutional Neural Network,Faster R-CNN）在SAR舰船目标检测的精度和速度两方面进行对比实验;最后在YOLOv5模型的基础上对主干网络（Backbone...     

星载合成孔径雷达探测水面舰船技术研究

星载合成孔径雷达(SAR)对水面舰船探测技术作为一种受到普遍关注的获取舰船信息的有效手段，较之其他探测方法有着独特的优点，根据近年来国内外运用星载SAR所取得的成果，对其探测舰船及其尾迹技术进行研究，以期获得较好的方法检测船舶及其尾迹。     

基于显著性分析的SAR溢油图像中暗斑区域检测方法研究

为了提高合成孔径雷达(synthetic aperture radar,SAR)图像中海上溢油的检测精度,设计了一种图像显著性和均值偏移分割相结合的暗斑检测方法,验证了该方法在SAR溢油图像暗斑识别中的有效性,为溢油区域的准确检测奠定了基础.针对SAR溢油图像中暗斑区域在人类视觉中属于感兴趣区域的特性,利用图像显著性检...     

一种基于箱线图的SAR图像舰船检测算法研究

将统计学中的箱线图应用于SAR图像舰船目标检测,提出了一种新的舰船检测算法.该算法不对海面背景建模,因此对复杂的海面背景具有较强的适应性,且能有效减少虚警和保持舰船轮廓.文中从适用条件、算法参数和时间复杂度三个方面,将该算法与恒虚警率(CFAR)算法做了详细的分析与比较,并推导出当海面背景符合高斯分布时,本文算法中的参...     

合成孔径雷达(SAR)图像的真实值定标与校验

随着合成孔径雷达（ＳＡＲ）的应用研究程度的深入，要求较为精确地掌握目标体的雷达后向散射特征，这样必须对ＳＡＲ图象进行真实值的定标与校验。本文基于基本散射理论对ＳＡＲ系统的增益进行分解，指出各个相关参数的获取方法，最后给出定标和数据处理流程图     

基于衰落统计特性的SAR图像仿真方法分析

合成孔径雷达(SAR)图像的斑点噪声严重阻碍了其数据的解译和应用,从数学物理的观点描述SAR图像的衰落统计特性,并将仿真结果与真实SAR图像作比较,得出两者具有一致性,最后提出根据SAR图像衰落统计特性的目标检测方法。     

合成孔径雷达影像反演台风浪的研究

台风浪的研究对于船舶航行、避风以及港口、海洋和近岸建筑物的安全有着重要的现实意义.本文选用台风“莫拉克”(MORAKOT)发生时的ERS-2卫星的合成孔径雷达(SAR)海面影像为例进行了相关研究,采用遥感软件和Matlab编程相结合,得到了台风浪的主波波长、波向和周期.与同实测资料吻合较好的Mike21软件的数值模拟结果相比较,效果良好.     

卫星SAR图像的海洋大气边界层特征参数的反演

针对中国南海ERS-2／SAR三维对流涡旋图像,分别利用Young的协方差方法和惯性子域方法,反演了对流海洋大气边界层高度、海洋大气边界层垂直对流尺度速度、表面浮力通量、Obukhov长度、稳定性校正后的海面风速和海面风速稳定性校正因子。结果表明,2种反演方法的结果符合较好,在一定程度上说明利用协方差方法和惯性子域方法从卫星SAR图像反演海洋大气边界层参数的有效性。     

一种舰船轴频电场实时检测方法

为了能够在低信噪比的条件下提高目标的检测概率,降低目标的漏警率和虚警率,提出了一种基于信号特征的舰船轴频电场实时检测方法。 根据舰船的螺旋桨转速与轴频的对应关系确定了特征频段, 将特征频段内经过线谱、连续谱分离及线谱识别后的功率谱估计的最大值作为特征量。 根据环境变化,设置具有自适应功能的动态阈值,对信号进行实时检测,确定疑似目标点。 然后通过模量判断,线谱倍频判断和线谱稳定性判断,最终确认目标是否存在。 对仿真信号和实测信号进行了特征提取和检测,在不同海况和较低信噪比下实现了目标检测。     

基于合成孔径雷达回波信号的海洋溢油监测方法研究

海洋油污染是各类海洋污染中最常见、分布面积最广且危害程度最大的污染之一。近年来,海洋特别是近海人类活动频繁,且随着海上运输和石油加工业的发展,油田井喷、钻井平台爆炸、船舶碰撞等所造成的溢油事故增多,因而,监测海洋溢油具有重要的经济和社会现实意义。研究采用MatLAB工具,通过图像预处理(图像校正和增强)、特征提取和神经网络识别等方法,对合成孔径雷达(SAR)海洋溢油图像进行处理,最终期望实现半自动区分SAR图像上各类目标,并进行多种神经网络方法效果比较。研究首先对SAR海洋溢油图像进行初步人工识别;然后进行图像预处理(几何校正、滤波处理等)和基于灰度共生矩阵的特征值计算;最后,借助神经网络方法对溢油区域和疑似溢油区域进行分类,输出分类处理后的图像。通过输出图像分析发现,神经网络能对SAR海洋溢油图像中溢油、海水、土地3类目标进行明确分类,且RBF神经网络模型精度高于BP神经网络。本文提出的半自动分类方法不仅能提高SAR图像处理效率,将分类目标扩充有溢油和非溢油扩充到溢油、海水、土地3类,提高图像处理的全面性,同时通过比较RBF和BP神经网络在SAR溢油图像分类上的具体优劣,有着较好实际意义。     

一种新的用于极化SAR图像船只检测的散射相似性测度

本文提出了一种新的用于极化合成孔径雷达（SAR）海上船只目标检测的测度。首先利用散射相似性参数研究船只与海杂波散射机制的差异。基于这些差异,提出了一种新的船只检测测度,该测度能够有效区分船只目标和海杂波。然后提出了利用核密度估计（KDE）方法对该测度进行建模的方法。基于统计模型,实现了自适应恒虚警率（CFAR）的检测方案。本文利用多景C波段RADARSAT-2极化SAR（Pol-SAR）数据上进行实验,系统分析了所提出测度的海杂波拟合性能与船只检测性能,并与两个经典的极化测度进行了比较,实验和比较结果证明了所提的测度的可行性。     

A new CFAR ship target detection method in SAR imagery

Many ship target detection methods have been developed since it was verified that ship could be imaged with the space-based SAR systems. Most developed detection methods mostly emphasized ship detection rate but not computation time. By making use of the advantages of the K-distribution CFAR method and two-parameter CFAR method, a new CFAR ship target detection algorithm was proposed. In that new method, we use the K-distribution CFAR method to calculate a global threshold with a certain false-alarm rate. Then the threshold is applied to the whole SAR imagery to determine the possible ship target pixels, and a binary image is given as the preliminary result. Mathematical morphological filter are used to filter the binary image. After that step, we use the two-parameter CFAR method to detect the ship targets. In the step, the local sliding window only works in the possible ship target pixels of the SAR imagery. That step avoids the statistical calculation of the background pixels, so the method proposed can much improve the processing speed. In order to test the new method, two SAR imagery with different background were used, and the detection result shows that that method can work well in different background circumstances with high detection rate. Moreover, a synchronous ship detection experiment was carried out in Qingdao port in October 28, 2005 to verify the new method and one ENVISAT ASAR imagery was acquired to detect ship targets. It can be concluded from the experiment that the new method not only has high detection rate, but also is time-consuming, and is suitable for the operational ship detection system.     

Semisupervised heterogeneous ensemble for ship target discrimination in synthetic aperture radar images

Ship detection using synthetic aperture radar (SAR) plays an important role in marine applications. The existing methods are capable of quickly obtaining many candidate targets, but numerous non-ship objects may be wrongly detected in complex backgrounds. These non-ship false alarms can be excluded by training discriminators, and the desired accuracy is obtained with enough verified samples. However, the reliable verification of targets in large-scene SAR images still inevitably requires manual interpretation, which is difficult and time consuming. To address this issue, a semisupervised heterogeneous ensemble ship target discrimination method based on a tri-training scheme is proposed to take advantage of the plentiful candidate targets. Specifically, various features commonly used in SAR image target discrimination are extracted, and several acknowledged classification models and their classic variants are investigated. Multiple discriminators are constructed by dividing these features into different groups and pairing them with each model. Then, the performance of all the discriminators is tested, and better discriminators are selected for implementing the semisupervised training process. These strategies enhance the diversity and reliability of the discriminators, and their heterogeneous ensemble makes more correct judgments on candidate targets, which facilitates further positive training. Experimental results demonstrate that the proposed method outperforms traditional tri-training.     

基于Radon变换的SAR图像船迹检测研究

给出了一种检测星载(合成孔径雷达SAR)图像中舰船尾迹的方法,该方法主要特点是将Radon变换局部化,即在对像素灰度值积分过程中沿着线性特征被分割后的若干短线段进行积分,而非对整张图像进行积分。该方法克服了Radon变换方法的局限性,使尾迹信号在背景噪声较强时依旧能被检测出,并且能检测出相对于整幅图像的尺寸较短小的尾迹。     

Extension of synthetic aperture radar (SAR) technique to undersea applications

This paper presents results on the extension of synthetic-aperture radar (SAR) technique to underwater mapping applications. Due to the modest propagation speed of acoustic waves in water, the areal-mapping rate is found to be limited. A number of approaches are discussed in the paper which have the potential of significantly increasing the mapping rate. The map resolutions, both in azimuth and range, are derived for a synthetic array system employing multiple elements. The digital processing requirements, such as the complex multiplication rate and memory size are determined. It was found that the digital signal-processing requirements for many viable mapping systems are within the capability of existing processing hardware.     

一种基于经验模态分解的SAR图像溢油识别算法

溢油事件的发生会给海洋环境的保护和经济发展带来巨大的影响。运用现代化的监测手段和技术进行监测,及时发现溢油现象和违规行为,保护海洋环境是非常重要的。合成孔径雷达（SAR）技术是溢油检测的有效工具,在SAR图像中溢油表现为黑色的区域,但是黑色区域也可能会由其他的因素引起。本文提出了一种基于二维经验模态分解（BEMD）的方法来识别溢油和疑似溢油。首先通过BEMD方法将感兴趣的区域分解为局部窄带的各分量—内蕴模函数（BIMF）之和,并对分解后得到的各分量IMF进行Hilbert变换,通过Hibert谱分析得到64维的特征空间,然后使用Relief方法得到5个特征向量,最后利用马氏距离分类器进行分类。通过实验结果表明,该方法能够有效、准确地检测出溢油,准确率超过90%。     

A synthetic aperture radar sea surface distribution estimation by n-order Bézier curve and its application in ship detection

To dates,most ship detection approaches for single-pol synthetic aperture radar(SAR) imagery try to ensure a constant false-alarm rate(CFAR).A high performance ship detector relies on two key components:an accurate estimation to a sea surface distribution and a fine designed CFAR algorithm.First,a novel nonparametric sea surface distribution estimation method is developed based on n-order Bézier curve.To estimate the sea surface distribution using n-order Bézier curve,an explicit analytical solution is derived based on a least square optimization,and the optimal selection also is presented to two essential parameters,the order n of Bézier curve and the number m of sample points.Next,to validate the ship detection performance of the estimated sea surface distribution,the estimated sea surface distribution by n-order Bézier curve is combined with a cell averaging CFAR(CA-CFAR).To eliminate the possible interfering ship targets in background window,an improved automatic censoring method is applied.Comprehensive experiments prove that in terms of sea surface estimation performance,the proposed method is as good as a traditional nonparametric Parzen window kernel method,and in most cases,outperforms two widely used parametric methods,K and G0 models.In terms of computation speed,a major advantage of the proposed estimation method is the time consuming only depended on the number m of sample points while independent of imagery size,which makes it can achieve a significant speed improvement to the Parzen window kernel method,and in some cases,it is even faster than two parametric methods.In terms of ship detection performance,the experiments show that the ship detector which constructed by the proposed sea surface distribution model and the given CA-CFAR algorithm has wide adaptability to different SAR sensors,resolutions and sea surface homogeneities and obtains a leading performance on the test dataset.     

Kelvin尾迹SAR多视向的成像仿真

利用船只Kelvin航迹模型、海面波模型和二尺度微波散射模型,提出了船尾迹多视向的成像仿真技术,并首次在二维空间中从不同视向仿真船尾迹的SAR图像。结果表明,当雷达视向与船只航向平行时,横波成像明显;当雷达视向与船只航向垂直时,扩散波成像明显;当雷达视向与船只航向有个夹角时,会出现一臂亮一臂暗的现象,这一现象取决于两臂尖波的传播方向与雷达视向的夹角,传播方向与雷达视向越接近平行的波越容易被雷达观测到,从而形成亮臂。仿真结果还得出另外一个结论:船只航向与雷达视向越接近垂直,两臂张角越小。仿真结果和实际的多幅ERS-SAR图像所观测到的结果是一致的。该模型可以有效地模拟Kelvin尾迹SAR多视向成像。     

Ship hull characteristics from surface wake synthetic aperture radar (SAR) imagery

Improved data collection and processing technologies along with the use of high resolution spectral techniques soon will make it possible to obtain estimates of the Kelvin wave amplitude function A(θ), ship speed U, and ship heading α from synthetic aperture radar images of ship wakes. This paper presents a series of methods for deriving additional hull characteristics such as the length L, volume V, and offsets ζ(x,z) from this spectral and surface wave information. The first method estimates the ship length by taking the Fourier transform of the slope amplitude function |kA|. The remaining estimates make use of the hull inversion code developed at the University of Michigan by Wu in 1991. The accuracy of the hull offsets predicted by the code is first determined for various options for solving the linear inversion problem. In this case, both the magnitude and phase of A(θ) are known in addition to the hull draft H. Since the draft is not often known a priori, the accuracy of the code is determined next by predicting the volume of the ship for an approximate though plausible input value of H. Finally, the accuracy of the non-linear inversion problem of obtaining offsets along the entire hull is investigated when only the magnitude but not the phase of A(θ) is known.