一种 ReliefF 和随机森林模型组合的多波束 海底底质分类方法

海底底质的快速探测和精细划分对海洋工程建设 、海洋资源开发等具有重要意义。多波束探测是目前声学底质遥测 的有效手段之一, 通常提取多波束反向散射强度图像和地形数据中的多维特征结合分类器进行底质分类。一方面, 若特征空 间维数过高, 分类效率会显著降低; 另一方面, 个别特征容易放大原始数据处理过程中仍存留的异常现象。针对这一问题, 本文提出了一种结合 Re1iefF 算法和随机森林 (Random Forest, RF) 算法的多波束底质分类方法。提取反向散射强度和地形 共 16 维特征, 利用Re1iefF 算法进行特征筛选, 排除低相关性特征, 降低特征空间维数, 结合采样点数据进行模型训练以构 建多波束底质分类模型。试验结合随机森林算法对未经特征筛选 、经主成分分析 (Principa1 Component Ana1ysis, PCA) 特征 优化后的特征进行分类实验作为对比。本文方法 Kappa 系数达到 85%, 分类总精度高于 90%, 精度具有明显优势, 耗时也 比较短。可见, 本文提出的结合 Re1iefF 和随机森林模型的多波束底质分类方法可以在保证分类精度的同时对多维特征进行 优化, 有效地提高了分类效率, 可对海底底质分类研究提供参考。     

基于多维度声学特征优选的多波束海底底质分类

海底表层底质分布信息的准确获取在构建海洋基础地理数据库中发挥着重要作用。目前,多波束是实现大范围海底底质分类的有效手段之一,基于多波束测深和反向散射强度数据所派生的声学特征被广泛应用于底质分类建模。然而,随着特征维度的增加,特征空间中存在的无关和冗余特征严重影响底质分类精度。为了定量评估声学特征对底质类别的表征能力,并消除无效特征对分类结果的干扰,本文提出了基于多维度声学特征优选的海底底质分类方法。首先,结合实际底质样本的物理属性对多维特征进行排序和优选,排除冗余和无关特征。其次,分别应用支持向量机、随机森林和深度信念网络构建海底底质监督分类模型。通过利用爱尔兰海南部多波束调查数据和实地取样信息进行试验,结果表明提出方法对海底底质的总体分类精度和Kappa系数分别最高达到了86.20%和0.834,相较于主成分分析和熵指标特征选择方法有明显提高,突出了该方法在海底底质探测及制图的应用潜力。     

基于测深数据的胶州湾底质类型估计方法

探讨了基于高精度多波束水深数据的底质分类方法。对高精度水深数据按一定采样窗口单元提取统计特征;利用聚类分析方法对采样窗口单元进行分类;将分类结果与表层沉积物底质调查结果以及声纳分类结果相对比,发现三者具有一致性。此方法可以用来识别基岩、砾石、沙和粘土等底质类型。     

基于改进BP神经网络的海底底质分类

通过采用遗传算法优化神经网络初始权值的方法,将GA算法与BP神经网络有机结合,应用于海底底质分类。基于多波束测深系统获取的反向散射强度数据,应用改进的BP神经网络分类方法,实现对海底基岩、砾石、砂、细砂和泥等底质类型的快速、准确识别。通过实验比较,GA-BP神经网络分类精度明显高于BP神经网络,证明了该方法的有效性和可靠性。     

用于海底底质分类的多波束声强数据选取研究

高质量的海底声强图是进行多波束海底底质分类、目标识别的基础。要得到"单纯"反映海底底质信息的声强图,就需要对原始声强数据进行地形改正,消除地形因素的影响。在描述了多波束数据中水深数据不能满足声强数据的改正要求问题的基础上,提出了以水深数据覆盖范围为约束的声强数据选取方法。实例计算结果表明:该方法在能有效地选取高质量的声强数据,提高了基于声强图像的海底底质分类精度。     

介绍几种典型的海底底质分类技术

本文主要介绍了四家海洋仪器公司的最新海底底质分类技术。它们分别利用多波束的反向散射强度数据、单波束的回声波形结构数据和旁侧声纳数据。采用了多参数统计分析、波形结构分析和影像属性分析等方法，实现了快速、高效、大面积地对海底底质进行间接的分类。尽管它们的技术各不相同，但都可以分为监督分类方式和非监督分类方式。     

基于SonarWiz的多波束声纳图像智能底质分类技术研究

多波束回波强度信息与海底底质类型具有较强的相关性,通过海底声纳图像能够实现底质类型的划分.为提高海底底质分类质量,依托SonarWiz的智能底质分类优势,在海底声纳图像纹理特征自动分类基础上,引入地形属性信息修正分类结果.以三亚崖州湾附近海域为例,基于实测海底地形数据和海底表面声纳图像,利用数据处理技术和图像分类方法,...     

海底声学底质分类的ELM-AdaBoost方法

基于自适应增强算法（AdaBoost）结合极限学习机（ELM）,通过迭代、调整、优化ELM分类器之间的权值,从而构建了具有强鲁棒性、高精度的ELM-AdaBoost强分类器,增强了现有的ELM分类器的稳定性。以珠江口海区侧扫声呐图像为实验数据,对礁石、砂、泥3类典型底质进行分类识别,该方法的平均分类精度超过90%,优于单一ELM分类器的平均分类精度85.95%,也优于LVQ、BP等传统分类器,且在分类所耗时间上也远少于传统分类器。实验结果表明,本文构建的ELM-AdaBoost方法可有效应用于海底声学底质分类,可满足实时底质分类的需求。     

济州岛南部海域海底声呐图像分析与声学底质分类

东海北部外陆架靠近济州岛南部海域,是黄海槽向冲绳海槽延伸的部分,属于黑潮分支黄海暖流的通道入口,分布着脊槽相间的海底底形,对其海底声呐图像的处理分析及声学底质分类的分析研究,有助于了解该通道海底底形表层纹理特征及沉积物分布规律。基于在济州岛南部海域获取的多波束声呐数据,应用图像处理技术和方法,对数据进行了处理,获得了海底声呐影像图,并对其表层纹理特征进行了描述和分析;同时,基于多波束反向散射强度数据,结合19组海底地质取样数据,建立研究区海底反向散射强度与沉积物粒度特征之间的统计关系模型,并以改进的学习向量量化神经网络方法,实现对海底粉砂质砂、黏土质砂以及砂-粉砂-黏土3种底质类型的快速自动分类识别。     

基于粒子群优化算法的PSO-BP海底声学底质分类方法

利用粒子群优化算法（PSO）较强的鲁棒性和全局搜索能力等优点,将PSO算法与BP神经网络相结合,优化了BP神经网络分类时的初始权值和阈值。基于珠江河口三角洲的侧扫声呐图像数据,提取了海底声呐图像中砂、礁石、泥3类典型底质的6种主要特征向量,利用PSO-BP方法对海底底质进行分类识别。实验表明,3类底质分类精度均大于90%,高于BP神经网络70%左右的分类精度,表明PSO-BP方法可有效应用于海底底质的分类识别。     

多波束反向散射强度数据处理研究

在探讨多波束测深系统反向散射强度与海底底质类型的关系基础上,研究影响反向散射强度的各种因素,主要分析了海底地形起伏、中央波束区反射信号对反向散射强度的影响,并给出了消除这些影响的方法;将处理后的“纯”反向散射强度数据镶嵌生成海底声像图,为海底底质类型划分以及地貌解译提供了基础数据和辅助判读依据.     

Processing Multibeam Backscatter Data

A new highly precise source of data has recently become available using multibeam sonar systems in hydrography. Multibeam sonar systems can provide hydrographic quality depth data as well as high-resolution seafloor sonar images. We utilize the seafloor backscatter strength data of each beam from multibeam sonar and the automatic classification technology so that we can get the seafloor type identification maps. In this article, analyzing all kinds of error effects in backscatter strength, data are based on the relationship between backscatter strength and seafloor types. We emphasize particularly analyzing the influences of local bottom slope and near nadir reflection in backscatter strength data. We also give the correction algorithms and results of these two influent factors. After processing the raw backscatter strength data and correcting error effects, we can get processed backscatter strength data which reflect the features of seafloor types only. Applying the processed backscatter strength data and mosaicked seafloor sonar images, we engage in seafloor classification and geomorphy interpretation in future research.     

Processing Multibeam Backscatter Data

A new highly precise source of data has recently become available using multibeam sonar systems in hydrography. Multibeam sonar systems can provide hydrographic quality depth data as well as high-resolution seafloor sonar images. We utilize the seafloor backscatter strength data of each beam from multibeam sonar and the automatic classification technology so that we can get the seafloor type identification maps. In this article, analyzing all kinds of error effects in backscatter strength, data are based on the relationship between backscatter strength and seafloor types. We emphasize particularly analyzing the influences of local bottom slope and near nadir reflection in backscatter strength data. We also give the correction algorithms and results of these two influent factors. After processing the raw backscatter strength data and correcting error effects, we can get processed backscatter strength data which reflect the features of seafloor types only. Applying the processed backscatter strength data and mosaicked seafloor sonar images, we engage in seafloor classification and geomorphy interpretation in future research.     

Seafloor acoustic remote sensing with multibeam echo-sounders and bathymetric sidescan sonar systems

This paper examines the potential for remote classification of seafloor terrains using a combination of quantitative acoustic backscatter measurements and high resolution bathymetry derived from two classes of sonar systems currently used by the marine research community: multibeam echo-sounders and bathymetric sidescans sonar systems. The high-resolution bathymetry is important, not only to determine the topography of the area surveyed, but to provide accurate bottom slope corrections needed to convert the arrival angles of the seafloor echoes received by the sonars into true angles of incidence. An angular dependence of seafloor acoustic backscatter can then be derived for each region surveyed, making it possible to construct maps of acoustic backscattering strength in geographic coordinates over the areas of interest. Such maps, when combined with the high-resolution bathymetric maps normally compiled from the data output by the above sonar systems, could be very effective tools to quantify bottom types on a regional basis, and to develop automatic seafloor classification routines.     

Seafloor Classification of Multibeam Sonar Data Using Neural Network Approach

In this study, the self-organizing map (SOM), which is an unsupervised clustering algorithm, and a supervised proportional learning vector quantization (PLVQ), are employed to develop a combined method of seafloor classification using multibeam sonar backscatter data. The PLVQ is a generalized learning vector quantization based on the proportional learning law (PLL). The proposed method was evaluated in an area where there are four types of sediments. The results show that the performance of the proposed method is better than the SOM and a statistical classification method.     

Seafloor Classification of Multibeam Sonar Data Using Neural Network Approach

In this study, the self-organizing map (SOM), which is an unsupervised clustering algorithm, and a supervised proportional learning vector quantization (PLVQ), are employed to develop a combined method of seafloor classification using multibeam sonar backscatter data. The PLVQ is a generalized learning vector quantization based on the proportional learning law (PLL). The proposed method was evaluated in an area where there are four types of sediments. The results show that the performance of the proposed method is better than the SOM and a statistical classification method.     

Remote estimation of surficial seafloor properties through the application Angular Range Analysis to multibeam sonar data

The variation of the backscatter strength with the angle of incidence is an intrinsic property of the seafloor, which can be used in methods for acoustic seafloor characterization. Although multibeam sonars acquire backscatter over a wide range of incidence angles, the angular information is normally neglected during standard backscatter processing and mosaicking. An approach called Angular Range Analysis has been developed to preserve the backscatter angular information, and use it for remote estimation of seafloor properties. Angular Range Analysis starts with the beam-by-beam time-series of acoustic backscatter provided by the multibeam sonar and then corrects the backscatter for seafloor slope, beam pattern, time varying and angle varying gains, and area of insonification. Subsequently a series of parameters are calculated from the stacking of consecutive time series over a spatial scale that approximates half of the swath width. Based on these calculated parameters and the inversion of an acoustic backscatter model, we estimate the acoustic impedance and the roughness of the insonified area on the seafloor. In the process of this inversion, the behavior of the model parameters is constrained by established inter-property relationships. The approach has been tested using a 300 kHz Simrad EM3000 multibeam sonar in Little Bay, NH. Impedance estimates are compared to in situ measurements of sound speed. The comparison shows a very good correlation, indicating the potential of this approach for robust seafloor characterization.     

Shallow-water imaging multibeam sonars: A new tool for investigating seafloor processes in the coastal zone and on the continental shelf

Hydrographic quality bathymetry and quantitative acoustic backscatter data are now being acquired in shallow water on a routine basis using high frequency multibeam sonars. The data provided by these systems produce hitherto unobtainable information about geomorphology and seafloor geologic processes in the coastal zone and on the continental shelf.Before one can use the multibeam data for hydrography or quantitative acoustic backscatter studies, however, it is essential to be able to correct for systematic errors in the data. For bathymetric data, artifacts common to deep-water systems (roll, refraction, positioning) need to be corrected. In addition, the potentially far greater effects of tides, heave, vessel lift/squat, antenna motion and internal time delays become of increasing importance in shallower water. Such artifacts now cause greater errors in hydrographic data quality than bottom detection. Many of these artifacts are a result of imperfect motion sensing, however, new methods such as differential GPS hold great potential for resolving such limitations. For backscatter data, while the system response is well characterised, significant post processing is required to remove residual effects of imaging geometry, gain adjustments and water column effects. With the removal of these system artifacts and the establishment of a calibrated test site in intertidal regions (where the seabed may be intimately examined by eye) one can build up a sediment classification scheme for routine regional seafloor identification.When properly processed, high frequency multibeam sonar data can provide a view of seafloor geology and geomorphology at resolutions of as little as a few decimetres. Specific applications include quantitative estimation of sediment transport rates in large-scale sediment waves, volume effects of iceberg scouring, extent and style of seafloor mass-wasting and delineation of structural trends in bedrock. In addition, the imagery potentially provides a means of quantitative classification of seafloor lithology, allowing sedimentologists the ability to examine spatial distributions of seabed sediment type without resorting to subjective estimation or prohibitively expensive bottom-sampling programs. Using Simrad EM100 and EM1000 sonars as an example, this paper illustrates the nature and scale of possible artifacts, the necessary post-processing steps and shows specific applications of these sonars.     

Sea-floor classification using multibeam echo-sounding angular backscatter data: a real-time approach employing hybrid neural network architecture

The presently studied numerical model, e.g., composite roughness, is successful for the purpose of seafloor classification employing processed multibeam angular backscatter data from manganese-nodule-bearing locations of the Central Indian Ocean Basin. Hybrid artificial neural network (ANN) architecture, comprised of the self-organizing feature map and learning vector quantization (LVQ), has been implemented as an alternative technique for sea-floor roughness classification, giving comparative results with the aforesaid numerical model for processed multibeam angular backscatter data. However, the composite-roughness model approach is protracted due to the inherent need for processed data including system-gain corrections. In order to establish that tedious processing of raw backscatter values is unessential for efficient classification, hybrid ANN architecture has been attempted here due to its nonparametric approach. In this technical communication, successful employment of LVQ algorithm for unprocessed (raw) multibeam backscatter data indicates true real-time classification application.     

Seafloor multibeam backscatter calibration experiment: comparing 45°-tilted 38-kHz split-beam echosounder and 30-kHz multibeam data

Obtaining absolute seafloor backscatter measurements from hydrographic multibeam echosounders is yet to be achieved. We propose a low-cost experiment to calibrate the various acquisition modes of a 30-kHz Kongsberg EM 302 multibeam echosounder in a range of water depths. We use a 38-kHz Simrad EK60 calibrated fisheries split-beam echosounder mounted at 45° angle on the vessel’s hull as a reference for the calibration. The processing to extract seafloor backscatter from the EK60 requires bottom detection, ray tracing and motion compensation to obtain acceptable geo-referenced backscatter measurements from this non-hydrographic system. Our experiment was run in Cook Strait, New Zealand, on well-known seafloor patches in shallow, mid, and deep-water depths. Despite acquisition issues due to weather, our results demonstrate the strong potential of such an approach to obtain system’s absolute calibration which is required for quantitative use of backscatter strength data.