【作者】 王慧；

【导师】 周卫东；

【作者基本信息】 哈尔滨工程大学 ， 导航、制导与控制， 2016， 博士

【摘要】 全天候、多方位立体的海洋动力环境监测对人类海洋活动、海洋探索和研究有着至关重要的推动作用,在众多先进设备中航海雷达以其便捷、投资小、高分辨率等特点成为国内外学者研究的热点。现阶段利用航海雷达采集的海杂波图像基于相应算法可提取多种海洋动力学参数,实现了其他遥感设备无法比拟的实时海面状态监测。本文结合实验室在研项目,基于航海雷达开展了海面风向提取技术的研究,为海洋动力环境监测提供了更加全面和精准的海面风向信息。本文基于课题的研究背景和意义,总结了国内外航海雷达海洋遥感的相应成果,并对基于航海雷达提取海面风向的关键技术进行了深入的研究,利用海面风场成像的小尺度风条纹特征提出了两套完整的海面风向反演技术方案,应用实测雷达图像序列对两种反演技术进行了测试和分析。1、针对雷达图像径向衰减问题,根据雷达图像径向非线性衰减特征,设计了基于最小二乘法自适应分段拟合算法校正雷达图像。应用最小二乘法分段拟合雷达径向回波强度,利用拟合函数值与真实值的绝对值误差作为自适应判断条件,实现了算法自动校正的能力。算法应用多元函数极值求解拉格朗日函数的最小二乘回归系数,保证了分段拟合的连续性,并且采用的最小二乘拟合函数最高阶为二阶,既简化了拟合函数模型又提高了算法的运行速度。结果显示拟合函数值与雷达图像径向回波强度偏差为0.83,标准误差22.19,相关性0.9986,符合拟合校正要求,证明了该相对辐射校正算法的有效性。2、海面风场成像特征的研究是应用航海雷达反演海面风向的基础,现阶段都是基于SAR成像的风条纹基础上进行研究的,但航海雷达与SAR在工作原理、测量方法上有较大差异。本文结合SAR成像机制,从海面风场调制对航海雷达成像的影响入手,理论分析了海面风场导致航海雷达成像特征的关键原因。根据海面风场静态特性,从航海雷达图像序列中提取出小尺度风条纹,并应用实测数据验证小尺度风条纹尺度、周期等特性。3、针对传统局部梯度法无法适应小尺度风条纹尺度变化的问题,提出基于自适应缩减算法(adaptive reduced method,ARM)的海面风向提取技术。算法首先利用自适应缩减算子对图像进行空间处理,获得缩减后图像;然后,构建缩减后图像梯度方向概率直方图,由梯度方向计算得到的稳定系数判断得出最优缩减率;最终,将最优缩减后图像通过优化Sobel算子提取出海面风向,并根据单点海面风向象限去除反演结果的180°模糊问题。应用实测数据得到反演结果与参考风向相关系数为0.9956,标准差为7.62°,偏差为-1.04°,相对传统局部梯度法海面风向反演精度提高了58.3%,验证了改进算法的有效性。4、空间域图像处理技术不仅会破坏小尺度风条纹特征,还会使缩减后图像丢失一部分小尺度风条纹图像,降低海面风向反演精度。本文创新性的提出基于能量谱(Energy Spectrum Method,ESM)海面风向提取技术。该技术首先应用图像2-D傅里叶变换,将海面静态特征图像转为频域能量谱图;然后,根据小尺度风条纹尺度设计带通滤波器,将小尺度风条纹能量谱从海面静态特征图像能量谱中提取出来;最后,利用小尺度风条纹与风向关系,由小尺度风条纹能量谱反演得到海面风向信息。在后续的数据实验中,实测数据反演结果与参考风向相关系数达到了0.99,偏差为-0.64°,标准差为8.29°。5、为了测试两种海面风向反演算法遥测海面风场的有效性,利用福建省海坛岛真实海洋环境设计测试技术方案,进而开展基于航海雷达提取海面风向的工程实验。结果显示ESM海面风向反演精度相对ARM提高了45.7%,并且ESM风向反演结果与参考风向的相关系数达到0.98,标准差12.13°,偏差1.68°,各项指标均优于ARM海面风向反演算法,在平均运行时间上ESM相对ARM也缩短了7s左右。实验还分析了测量区域面积、海面风速、有效波高及流向对两种算法的影响,实验证明ESM相对ARM拥有更高的工程应用价值,但在实验情况下ARM算法具有应用简便的优势。更多还原

【Abstract】 The all-weather,all-round three-dimensional ocean dynamic environment monitoring has a vital role of human marine activities,marine exploration and research.Owing to convenience,small investment,and high resolution,navigation radar has become a hot topic in the study of scholars in overseas and domestic research status.At this stage,using the corresponding algorithm,we can derive the ocean dynamic parameters from the sea-clutter images collected by marine radar.This approach has an outstanding performance than other remote marine monitoring sensors.This thesis researches the technology of ocean wind-direction retrieval according to project of marine lab,and dedicates to giving comprehensive and precise ocean-wind information.This thesis reviewed the project significance and background,gave the summaries of radar ocean previous research results at home and abroad.Meanwhile,we gave an deep investigation on the technology of ocean wind-direction retrieval based on marine radar,designed two overall ocean wind-direction retrieval schemes using small scale wind streak characteristics.Finally,we test and verify the two different retrieval technology availability using real-time radar sequences.In order to remove the radar image radial attenuation,we design a least squares adaptive piecewise fitting algorithm to correct radar image according to radial attenuation characters.The algorithm application of least square piecewise fitting radar radial echo intensity,achieves adaptive judgment using the absolute error of fitting function value and real value.This algorithm uses multivariate function extremum to solve least-squares regression coefficient of Lagrange function,hence guarantees the continuity of piecewise fitting.The least squares using the second order as the highest order,means this approach can not only simplify the fitting function model but also improve the algorithm running speed.Results show that the fitting function value and radar radial echo intensity deviation is 0.83°,standard error is 22.19°,and correlation coefficient is 0.9986,matches the requirements of the radar correction.It verifies the effectiveness of the proposed relative radiation correction algorithm.Research of wind streak imaging mechanism is the foundation to retrieve ocean wind direction,but all the research achievements are based on the wind streaks from SAR at percent stage.Navigation radar and SAR have a bigger difference in terms of working principle,measuring method,which will cause the algorithm incorrect.This thesis begin with the influence of sea surface wind field modulation for radar imaging,theoretically analysis the key reasons of marine radar imaging characteristics generated from sea surface wind field.According to the sea surface wind field static characteristics,we extract small scale wind streaks from navigation radar image sequences,and ultimately identify the scale,cycle and axial direction of small scale wind streaks using the measured radar data.To overcome the problem that conventional local gradient method cannot adapt the small scale wind streak changing,we proposed an adaptive reduction operator method(ARM)to extract wind direction from marine radar image.Firstly,this algorithm use adaptive reduction operator to processing image in space,then a stable coefficient can be derived from probability diagram using reduction gradient direction information.Finally,we can extract the wind direction using optimal Sobel operator from the reduction image,and eliminate the inversion 180°ambiguous according to the quadrant of single point value.The correlation coefficient between real-time inversion results and reference ocean wind direction is 0.9956,with a standard deviation 7.62°,a bias-1.04°,owning a 58.3% improvement compared to conventional approach.The space domain image processing technique can corrupt small scale wind streaks,as well as lose a part of image,leading to lower inversion accuracy.To address this problem,an energy spectrum method(ESM)ocean wind direction extraction is proposed.The image will be convert into frequency domain energy spectrum diagram using 2-D Fourier transform,then small scale wind streaks can be derived after a band filter.Finally,ocean wind direction information is reflected by small scale wind streaks energy spectrum.The correlation coefficient between real-time retrieval results and reference sea wind direction is 0.99,with a standard deviation 8.29°,a bias-0.64°,verify the algorithm effective.To test the effective of the prior proposed sea wind inversion algorithm,we design a real ocean experiment based on marine radar at Pingtan island of Fujian province.The results shows that the EMS has a 45.7% improvement compared ARM,and the EMS correlation coefficient achieve 0.98,with a standard deviation 12.13°,a bias 1.68°.All indexes are better than ARM ocean wind inversion algorithm.Further,7s the average operation time shorten 7s for EMS than ARM.This experiment meanwhile analyzed measurement area,ocean surface wind speed,significant wave height and wave current influence on the ARM and EMS algorithm.The results show that EMS has a better engineering significance compared to ARM,although ARM has a little advantage in several experiment cases.更多还原