【作者】 李英；

【导师】 沈继红；

【作者基本信息】 哈尔滨工程大学 ， 系统工程， 2013， 博士

【摘要】 海浪是一种常见的海洋波动现象，它与海洋探索、海洋开发、海洋军事技术和海上航行安全等都有着十分密切的关系。对海洋环境实现全方位、多手段的立体监测，可以使人类及时、准确地掌握海洋自身运动变化的规律，以满足军事和民用两方面的需求。X波段航海雷达以其高分辨率、使用便捷等优点逐渐成为海洋遥感领域的一个重要分支，也吸引了越来越多海洋遥感科学家的注意。因此，开展基于X波段航海雷达的海浪信息提取技术研究和设备开发具有重要的理论和现实意义。本文结合课题组在研项目，对X波段航海雷达海浪参数反演过程中的相关问题，如同频干扰、目标物和降雨等的噪声抑制，海流估算，带通滤波器设计和有效波高反演等进行了深入的研究，主要内容如下：1、提出了一种将拉普拉斯算子和改进拉普拉斯算子相结合的同频干扰检测方法。海杂波图像中存在一定的同频干扰噪声，已有的同频干扰抑制方法大都是基于硬件设计的，而基于软件的处理方法大都会在处理同频干扰的同时导致海浪信息的丢失。因此本文提出了针对同频干扰的检测方法，并利用基于反距离加权插值及基于三角网格插值相结合的方法对雷达图像进行插值修复。对上述方法进行实验，并与中值滤波、均值滤波结果进行对比，验证了所设计方法的有效性。2、利用Adaline神经网络估计图像回波强度分布模型的参数，提出了一种基于雷达图像时间序列回波信号时、空分布特性的目标物噪声检测方法。已有的目标物检测方法大都是基于单幅SAR图像展开的，目标物在图像中有清晰的边缘轮廓特征，而在本课题组雷达图像中，目标物的边缘是渐变的，和海浪信息没有明显的分界。为了适应课题的需要，设计该检测方法，并对上述方法进行实验验证，结果表明所设计的检测方法能快速有效的检测到回波图像中的目标物。3、提出了一种使用质量控制、表面纹理视在特性比、信噪比和回波强度均值四个参数相结合的降雨噪声检测方法。已有的降雨干扰抑制方法大都是基于硬件设计的，基于软件设计的方法主要是利用二维表面参数（均值和差异系数）进行检测然后进行滤波处理。在本实验中，二维参数并不能有效地识别降雨图像，因此本文提出了针对降雨干扰的检测方法，对降雨影响较轻的图像进行频域滤波处理，并将滤波结果与中值滤波的结果进行对比。针对以上方法进行实验，实验结果表明所设计方法能有效地识别降雨和未降雨图像，并能有效地抑制降雨干扰。4、提出了一种基于粒子群优化算法的海流信息提取方法以及基于基本色散关系和色散关系极限边界的带通滤波器。经验带通滤波器存在滤波信息不准的问题，需要对滤波器进行改进设计。由于海流在色散关系中起了重要的作用，而色散关系是带通滤波器的基础，所以带通滤波器的改进设计首要的是提高海流估算的准确度。本文介绍了几种目前常用的基于最小二乘原理的海流估算方法和几种经验带通滤波器，然后就所设计的估流方法和带通滤波器进行实验，实验结果验证了所设计方法的有效性。5、提出了一种基于最小圆半径的野值剔除和数据稀疏化方法以及基于PSO的分段线性化有效波高反演方法。首先介绍了传统的线性模型建模方法和基本的分段线性化模型建模方法，并在此基础上设计了基于PSO的分段线性化建模方法，详细地给出了基本原理和流程。对信噪比和有效波高的原始反演结果进行了基于最小圆半径的野值剔除和数据稀疏化处理，并利用经过处理的数据对几种建模方法进行实验。实验结果验证了数据预处理方法和基于PSO的有效波高反演模型建模方法的有效性。更多还原

【Abstract】 Ocean wave is a common ocean fluctuation phenomenon, and it has an importantinfluence on human activities about ocean, such as exploration, development, militarytechnology, safe navigation and so on. Making omnibearing and comprehensive-meansmonitor on the ocean environment enables human to understand the variation of ocean motiontimely and accurately, and it also meets the needs of both the military and the civilian aspects.With the advantages of high resolution and ease of use, the X-band marine radar has been animportant branch of ocean remote sensing gradually. And it has been attracting more and moreattentions of the ocean remote sensing scientists. Therefore, it has important theoretical andpractical significance to research wave information extraction technology and equipmentdevelopment based on the X-band marine radar.Based on the studied project of our research group, some key problems in the waveparameters inversion progress of X-band marine radar are deeply studied, such as the noisesuppression of co-channel interference, fixed target and rain, the ocean surface currentinversion, the band-pass filter design, the significant wave height inversion and so on. Themain contents are as follows:1．A noise detection method combined the Laplace operator with the improved Laplaceoperator is proposed. Suppression methods of co-channel interference are mostly based onhardware. And most of the software methods lose the wave information while progressing. Soa noise detection method is proposed, and a noise processing method based on anti-distanceweighted interpolation and triangulation grid interpolation are applied. Compared with theresults of median filter and mean filter, the experiment results show that the new methods areeffective.2．After the parameters of the distribution model estimated by Adaline neural network, adetection method based on the spatial and temporal distribution characteristics of the radarecho signals and with the radar time sequence images is proposed. Detection methods oftargets are mostly based on single SAR image. In SAR images, targets have clear edgecontour features, while the edges of the targets in radar images of this experiment changegradually, and do not have obvious boundaries. In order to meet the needs of our project, thispaper gives a detailed analysis of the spatial and temporal distribution characteristics of theradar echo signals, and designs a new detection method. The experiment results show that thenew method can detect the targets quickly and efficiently. 3．A rainfall identification method taking advantage of quality control, surface roughnessparameter, signal to noise ratio and echo intensity mean is proposed. Existing suppressionmethods of rain interference are mostly making use of hardware. And identification methodwith mean and difference coefficient of images can not effectively identify rainfall images inexperiments. So, on the basis of previous research results, the new identification method isdesigned. The noise images slightly affected by the rainfall are filtered in frequency domain.Experiment results show that the new method can identify and filter the rainfall interferenceefficiently.4． An extraction method of the ocean surface current based on Particle SwarmOptimization and a band-pass filter combined with both the fundamental mode dispersionrelation and the boundary of dispersion relation are proposed. Existing band-pass filters havethe shortcoming that the extracted information is not accurate, so the improvement is needed.The ocean surface current plays an important role in the dispersion relation, which is the basisof band-pass filter. So, the first task of band-pass filter improvement is to improve the oceansurface current estimation. Several currently used ocean current extraction methods based onthe principle of least squares method are introduced. Then some existing band-pass filters aredescribed. Experiment results show that these new methods can improve the accuracy ofextracted wave parameters efficiently.5．Methods of outliers rejection and data rarefaction based on the minimum radius and anew piecewise linear inversion method of the significant wave height based on PSO areproposed. The traditional linear modeling method and the basic piecewise linear modelingmethods are introduced, and the detailed basic principles and processes are illustrated. Thispaper proposes demand of outliers rejection and data rarefaction, also gives the correspondingmethods based on minimum radius. Finally experiments on these modeling methods withdealt data are carried out, and the results show that the designed methods are effective.更多还原