【作者】 陆丹；

【导师】 李海森；

【作者基本信息】 哈尔滨工程大学 ， 通信与信息系统， 2012， 博士

【摘要】 海底地形作为海洋环境的主要组成部分，在海洋开发领域具有重要价值。多波束测深技术是当代海底地形勘测的一项高新技术，目前已成为海洋监测与海底资源调查的最主要手段之一，被广泛应用于海道测量、海底资源勘测、海底目标探测、水上安全航行等众多领域。随着科技的不断发展与进步，目前多波束测深系统已实现超宽覆盖、高分辨测量，其采集的高密度测深数据已达到百万级、千万级乃至海量的数据量。作为多波束测深数据后处理技术的重要组成部分，大数据量测深数据的自动处理方法与海底成图技术是多波束测深数据可靠应用的重要途径，并且是长期以来多波束测深数据后处理研究中的热点与难点之一。本文结合当前国内外多波束测深数据后处理技术发展趋势和国产浅水宽覆盖多波束测深声呐研制实际需求，围绕测深数据后处理中的异常值自动检测与剔除方法、基于联合不确定度的多波束测深估计方法、大数据量海底数字地形建模与可视化技术以及海底地形等深线快速生成技术等四个方面展开研究，主要研究内容如下：首先，研究多波束测深异常值自动检测与剔除方法。在分析比较目前存在的三种主要测深异常值自动检测方法各自优缺点的基础上，提出一种基于截断最小二乘估计趋势面滤波的稳健异常值检测与剔除方法，并通过计算机仿真验证该方法对离散存在和簇群存在的异常值的检测性能。根据多波束测深数据点的分布特性，提出采用联合全局与局部方差的动态门限作为多波束测深数据的异常值检测门限，并通过计算机仿真验证该动态门限的有效性与适用性。对实际多波束测深数据进行异常值检测与剔除处理，验证了基于截断最小二乘估计趋势面滤波的异常值自动检测与剔除方法的有效性与实用性。其次，研究基于联合不确定度的多波束测深估计方法。深入分析影响多波束测深结果不确定度的各项因素，建立多波束测深的不确定度传播模型，获取实际多波束测深数据的水平不确定度与垂直不确定度。在获得每个测深点水平与垂直不确定度的基础上，进行基于联合不确定度的多波束测深估计研究，提出一种基于局部曲面拟合的节点深度预测方法以获得斜坡地形上测深点对节点深度估计的正确预测。对节点多重估计追踪与最优估计选取准则进行研究，提出一种基于局部最优均深度的最优估计选取准则。通过对海试数据的处理验证基于联合不确定度的多波束测深估计方法的高效性与稳健性。再次，研究大数据量海底地形快速建模与可视化技术。研究分析大数据量多波束测深数据的快速海底地形建模方法基础上，提出一种基于不完全二叉树的动态分块与合并机制，采用对测区数据点先进行分块构网再逆向合并的方法快速建立海底地形模型。针对子网合并过程中由离散点分布不确定性和合并算法复杂性引起的浮点计算误差错误发生机率增大而导致整个构网程序的稳健性降低的问题，提出一种联合矢积测试的双向缝合算法。通过实际多波束测深数据的处理验证联合矢积测试的双向缝合算法的高效性与稳健性。最后，研究等海底地形深线快速生成算法。根据大数据量多波束海底地形等深线快速生成与显示需求，提出一种基于等深值索引序列的等深线快速生成算法。通过将浮点型深度属性三角网转化为整型索引属性三角网，进一步提出一种基于0/1异或的等深线走向快速判定方法进行等深线跟踪，实现了等深线的快速生成。在分析比较三次Bezier函数法和三次B样条函数法这两种等深线平滑方法的基础上，采用三次B样条函数法实现等深线平滑。通过对实际多波束测深数据的处理，验证基于等深值索引序列的海底地形等深线快速生成算法的高效性与稳健性以及三次B样条函数等深线平滑方法的实用性。更多还原

【Abstract】 As a major component of marine environment, the seafloor terrain is important in theocean exploration. The Multi-beam bathymetric technique has become an up-to-datetechnology in contemporary seafloor topography survey and also plays a major part in oceanmonitoring and the investigation of seabed resources. It has been widely applied in manyfields, such as the hydrographic survey, the seabed resources investigation, the underwatertarget detection and the navigation, etc. With the development of science and technology, themultibeam bathymetry system has achieved super wide coverage measurement and highresolution measurement. Moreover, the amount of high-density bathymetric data collected hasreached the magnitude of one million, ten millions and even more. The automatic signalprocessing and the seafloor mapping on large amount of bathymetric data are importanttechniques in multibeam bathymetric data post-processing, which is an important and reliableapplication of the data and has been a hot yet a difficult research issue for a long time. Tomeet the demand of developing the wide-coverage multibeam bathymetry sonar in shallowwater in the country and also to follow the recent development of the techniques world widelyin multibeam bathymetric data post-processing, this thesis is devoted to four research aspects,namely the algorithm to automatically detect and eliminate the outliers in multibeambathymetric data, the multibeam bathymetry estimation based on the combined uncertainty,the modeling and visualization techniques in the seafloor digital terrain with large amount ofdata and the fast generation of the iso-depth contours. More specific contents are as follows.The first part of the thesis is devoted to developing the algorithm in automaticallydetecting and eliminating outlier in bathymetric data. After analyzing and the existing threemethods and comparing their advantages and disadvantages respectively, a robust automaticoutlier detection and elimination algorithm based on the trimmed least squares estimation isproposed. The detection performance of the algorithm for discrete outliers and the clusters hasbeen proved by computer simulation. According to the distribution characteristics ofmultibeam bathymetric data points, a dynamic threshold associated with the global and thelocal variance is proposed as the threshold for outlier detection, the applicability andefficiency of this dynamic threshold have been verified by computer simulations. After that,the outlier detection and elimination have been carried out using real multibeam bathymetricdata in the thesis and the results have confirmed this newly proposed method is effective andapplicable. The second part of this thesis has been devoted to the multibeam bathymetry estimationbased on the combined uncertainty. Various factors affecting the uncertainty of multibeambathymetric data are analyzed. Then, in order to obtain the horizontal and vertical uncertaintyof actual multibeam bathymetric data, the propagation models for multibeam bathymetricuncertainty are established. After that, the multibeam bathymetry estimation based on thecombined uncertainty is discussed using the horizontal and vertical uncertainty of each ofsounding. Following that, a node depth prediction method based on the local surface fitting isproposed to give the correct depth estimations for the nodes on the slopes. Moreover, in theend of this part of the thesis, the multiple estimation tracking are discussed and a robustoptimal estimation selecting rule is proposed based on the local average depth optimum. Theprocessing results using the data from sea trials have confirmed the robustness andeffectiveness of multibeam bathymetry estimation based on the combined uncertainty.The third part of the thesis is devoted to the fast modeling and visualization of theseafloor terrain. A dynamic block dividing and merging mechanism is proposed based on theincomplete binary tree after studying the fast modeling of the seafloor terrain using largeamount of multibeam bathymetric data. By firstly dividing the large amount of data in thesurvey area into several blocks and then constructing Delaunay triangulations in eachsub-block individually, the subtriangulations are merged in a regressive order to form aintegrated triangulation mesh. To deal with the low efficiency and low robustness oftriangulation procedure caused by both the uncertain distribution of sounding points and thecomplexity of the merging algorithm during the subtriangulation merging, towards the end ofthis part of the thesis, a bidirectional sew algorithm is proposed associated with the crossvector product test. The processing results of actual bathymetric data has confirmed that thebidirectional sew algorithm is effective and robust.The final part of the thesis is devoted to the fast algorithm in generating the iso-depthcontours in the seafloor. To efficiently generate the contours from large amount of seafloorterrain data, a fast algorithm of contours generation is proposed based on the indexingsequence of iso-depth values. The algorithm transforms the original triangulation net of floatdepth values to integer indexes. Following that, a fast decision-making method based on0/1exclusive-OR is proposed for telling the contour alignment, which finally accomplishescontours tracing and their fast generation. Moreover, the cubic B-spline function method isadopted for contour smoothing after the comparison with the cubic Bezier function method.By processing actual multibeam bathymetric data, the contours generation based on theindexing sequence of iso-depth values is proved to be effective and robust and the Cubic B-spline function method to be applicable.更多还原