学习向量量化神经网络在多波束底质分类中的应用研究

利用多波束测深系统获取的反向散射强度数据，应用学习向量量化（learning vector quantization，LVQ）神经网络分类方法实现了对海底砂、砾石和基岩等底质类型的快速、有效的识别。通过比较，证明了该方法能较好地区分出不同海底底质类型。     

海底底质分类反向散射强度三维概率密度法

反向散射强度随入射角变化信息蕴含海底底质特性,是多波束海底底质分类的重要数据源。目前,多波束海底底质分类主要提取平均反向散射强度随入射角的变化信息或基于声呐图像,没有综合考虑两者的相关信息。本文综合考虑平均反向散射强度随入射角变化信息及其概率密度分布,绘制三维概率密度图,提出了基于反向散射强度三维概率密度图的海底底质分类方法。实例计算结果表明:该方法能直观表达多波束条带覆盖区内不同海底底质类型的数量,有效判断不同海底底质类型的边界,实现海底底质类型的快速识别。     

利用深层卷积神经网络实现地形辅助的多波束海底底质分类

海底底质分类对于海洋资源开发与利用、海洋科学研究等多方面具有重要意义。目前,多波束探测是实现大范围海底底质分类的有效手段之一,通常基于多波束反向散射强度提取角度响应(AR)特征及反向散射图像特征进行底质分类。由于特征来源较单一,分类器结构简单,往往分类精度不高。为此,本文提出了一种基于深层卷积神经网络(CNN)的多波束海底底质分类方法。除反向散射强度特征外,还利用地形特征,将特征向量转换为波形图,再输入卷积神经网络进行训练和分类。试验对比不同特征组合以及BP网络、支持向量机(SVM)、K近邻(KNN)、随机森林(RF)4种常规分类器,本文模型算法总体分类精度达到94.86%,Kappa系数为0.93,精度具有明显优势,效率也比较高。表明该方法有效利用两种数据类型所蕴含的海底底质信息,充分发挥卷积神经网络权值共享、高效率等特点,实现高分辨率海底底质分类,可对海底底质分类研究提供参考。     

Simrad EM多波束反向散射强度数据精处理研究

本文在探讨多波束反向散射强度与海底底质类型关系的基础上,主要针对目前广泛使用的Simrad EM系列多波束,详细描述了在数据获取期间系统进行的实时补偿,分析了反向散射强度数据中残留的入射角和波束模式指向性等因素的影响,给出了相应的改正方法。计算结果表明,本文给出的方法可对反向散射强度数据进行有效的处理,为正确使用声纳图像进行海底底质的判读提供了保证。     

Seafloor sediment classification based on multibeam sonar data

The multibeam sonars can provide hydrographic quality depth data as well as hold the potential to provide calibrated measurements of the seafloor acoustic backscattering strength. There has been much interest in utilizing backscatters and images from multibeam sonar for seabed type identification and most results are obtained. This paper has presented a focused review of several main methods and recent developments of seafloor classification utilizing multibeam sonar data or/and images. These are including the power spectral analysis methods, the texture analysis, traditional Bayesian classification theory and the most active neural network approaches.     

Seafloor sediment classification based on multibeam sonar data

The multibeam sonars can provide hydrographic quality depth data as well as hold the potential to provide calibrated measurements of the seafloor acoustic backscattering strength. There has been much interest in utilizing backscatters and images from multibeam sonar for seabed type identification and most results are obtained. This paper has presented a focused review of several main methods and recent developments of seafloor classification utilizing multibeam sonar data or/and images. These are including the power spectral analysis methods, the texture analysis, traditional Bayesian classification theory and the most active neural network approaches.     

Backscatter coefficient as an attribute for the classification of full-waveform airborne laser scanning data in urban areas

Airborne laser scanning (ALS) data are increasingly being used for land cover classification. The amplitudes of echoes from targets, available from full-waveform ALS data, have been found to be useful in the classification of land cover. However, the amplitude of an echo is dependent on various factors such as the range and incidence angle, which makes it difficult to develop a classification method which can be applied to full-waveform ALS data from different sites, scanning geometries and sensors. Additional information available from full-waveform ALS data, such as range and echo width, can be used for radiometric calibration, and to derive backscatter cross section. The backscatter cross section of a target is the physical cross sectional area of an idealised isotropic target, which has the same intensity as the selected target. The backscatter coefficient is the backscatter cross section per unit area. In this study, the amplitude, backscatter cross section and backscatter coefficient of echoes from ALS point cloud data collected from two different sites are analysed based on urban land cover classes. The application of decision tree classifiers developed using data from the first study area on the second demonstrates the advantage of using the backscatter coefficient in classification methods, along with spatial attributes. It is shown that the accuracy of classification of the second study area using the backscatter coefficient (kappa coefficient 0.89) is higher than those using the amplitude (kappa coefficient 0.67) or backscatter cross section (kappa coefficient 0.68). This attribute is especially useful for separating road and grass.     

海底入射角对多波束反向散射强度的影响及其改正

综合考虑测船姿态、声线弯曲和海底地形对入射角的影响,推导了计算多波束海底入射角的实用模型。根据Lambert法则,改正海底入射角对反向散射强度的影响。实例计算结果表明,所述模型能精确地计算波束在海底的入射角,有效地改正其对反向散射强度的影响。     

多波束后向散射数据地形校正的算法研究

利用多波束反向散射数据进行海底底质分类是目前多波束声纳系统应用的一个热点研究方向.然而,由于海底不规则地形的影响,多波束声纳接收的反向散射信号,往往不能够真实表达海底底质的性质.因此,消除地形对多波束反向散射数据的影响是提高海底底质分类精度的一个重要步骤.本文基于多波束系统的特点,在详尽分析了地形对多波束反响散射数据影...     

Effect of Long Waves on Ku-Band Ocean Radar Backscatter at Low Incidence Angles Using TRMM and Altimeter Data

This letter uses a large ocean satellite data set to document relationships between Ku-band radar backscatter (sigma_o) of the sea surface, near-surface wind speed (U), and ocean wave height (SWH). The observations come from satellite crossovers of the Tropical Rainfall Mapping Mission (TRMM) Precipitation Radar (PR) and two satellite altimeters, namely: 1) Jason-1 and 2) ENVISAT. At these nodes, we obtain TRMM clear-air normalized radar cross-section data along with coincident altimeter-derived significant wave height. Wind speed estimates come from the European Centre for Medium-Range Weather Forecast. TRMM PR is the first satellite to measure low incidence Ku-band ocean backscatter at a continuum of incidence angles from 0deg to 18deg. This letter utilizes these global ocean data to assess hypotheses developed in past theoretical and field studies.     

海底大地测量控制网研究进展综述

高精度的海底大地测量控制网是数字海洋和水下网络中心战的基础设施之一,而现阶段我国海底控制网的建设和研究几乎为空白。为自主建立我国高精度的海底大地测量控制网,本文借鉴发达国家的经验,结合国外海底控制网的发展与研究现状,系统介绍了GNSS/Acoustic系统的组成及发展;详细梳理了数据处理中的关键技术,分析了影响海底控制点精度的因素和控制点定位精度;结合我国现状,建议加紧开展我国海底控制网的建设与研究。     

Evaluation of RADARSAT Standard Beam data for identification of potato and rice crops in India

The Canadian satellite RADARSAT launched in November 1995 acquires C-band HH polarisation Synthetic Aperture Radar (SAR) data in various incident angles and spatial resolutions. In this study, the Standard Beam S7 SAR data with 45°–49° incidence angle has been used to discriminate rice and potato crops grown in the Gangetic plains of West Bengal state. Four-date data acquired in the 24-day repeat cycle between January 2 and March 15, 1997 was used to study the temporal backscatter characteristics of these crops in relation to the growth stages. Two, three and four-date data were used to classify the crops. The results show that the backscatter was the lowest during puddling of rice fields and increased as the crop growth progressed. The backscatter during this period changed from −18 dB to −8 dB. This temporal behaviour was similar to that observed in case of ERS-SAR data. The classification accuracy of rice areas was 94% using four-date data. Two-date data, one corresponding to pre-field preparation and the other corresponding to transplantation stage, resulted in 92% accuracy. The last observation is of particular interest as one may estimate the crop area as early as within 20–30 days of transplantation. Such an early estimate is not feasible using optical remote sensing data or ERS-SAR data. The backscatter of potato crop varied from −9 dB to −6 dB during the growth phase and showed large variations during early vegetative stage. Two-date data, one acquired during 40–45 days of planting and another at maturing stage, resulted in 93% classification accuracy for potato. All other combinations of two-date data resulted in less than 90% classification accuracy for potato.     

A semilabeled-sample-driven bagging technique for ill-posed classification problems

In this letter, a semilabeled-sample-driven bootstrap aggregating (bagging) technique based on a co-inference (inductive and transductive) framework is proposed for addressing ill-posed classification problems. The novelties of the proposed technique lie in: 1) the definition of a general classification strategy for ill-posed problems by the joint use of training and semilabeled samples (i.e., original unlabeled samples labeled by the classification process); and 2) the design of an effective bagging method (driven by semilabeled samples) for a proper exploitation of different classifiers based on bootstrapped hybrid training sets. Although the proposed technique is general and can be applied to any classification algorithm, in this letter multilayer perceptron neural networks (MLPs) are used to develop the basic classifier of the proposed architecture. In this context, a novel cost function for the training of MLPs is defined, which properly considers the contribution of semilabeled samples in the learning of each member of the ensemble. The experimental results, which are obtained on different ill-posed classification problems, confirm the effectiveness of the proposed technique.     

A seafloor experiment to monitor vertical deformation at the Lucky Strike volcano, Mid-Atlantic Ridge

Decades of cruise-based exploration have provided excellent snapshots of the structure of mid-ocean ridges and have revealed that accretion is a mixture of steady-state and quantum events. Observatory-type studies are now needed to quantify the temporal evolution of these systems. A multi-disciplinary seafloor observatory site is currently being set up at the Lucky Strike volcano, in the axial valley of the slow spreading Mid-Atlantic ridge as a part of the MoMAR (monitoring of the Mid-Atlantic Ridge) initiative. The aim of this observatory is to better understand the dynamics of the volcano and the hydrothermal vents hosted at its summit as well as their plumbing systems. In August 2006, the GRAVILUCK cruise initiated an experiment to monitor the deformation of Lucky Strike volcano. A geodetic network was installed, and seafloor pressure, gravity and magnetic data were collected. In this paper, we present the method used to monitor volcanic deformation, which involves measuring relative depth difference between points within a seafloor geodesy network. We show that, taking into account oceanographic variability and measurement noise, the network should be able to detect vertical deformations of the order of 1 cm.     

基于目标时域散射特性的土地覆盖类型分类研究

目标散射特性随时间变化的规律称之为目标时域散射特性。目标时域散射特征是利用多时相雷达遥感图像进行目标识别的基础。本研究以广东肇庆为试验区,利用多时相单参数雷达图像进行土地覆盖类型分类研究,分析了试验区内典型植被的结构、形态及其散射机理等特征,总结了各类目标的时域散射特性,区分识别了多种目标,制作了土地覆盖类型分类图.     

加权最小二乘估计的多波束声强数据归一化方法

高质量的海底声强图是多波束海底底质分类的基础。在系统分析了多波束声强图像中沿测船航迹灰度异常值和区域性明暗差异成因的基础上,提出了基于加权最小二乘估计的多波束声强数据归一化方法。实例计算结果表明:该方法在不损失原有强度变化信息的基础上,有效地抑制了声强图中灰度异常值对海底底质分类的影响,提高了区域底质分类的质量。     

GNSS‐声学海底定位的声速误差处理方法综述

全球导航卫星系统（global navigation satellite system,GNSS）‐声学海底定位是面向海底俯冲带板块形变监测需求提出的一种定位技术,也是建设海洋时空基准网的一种重要技术,有着广阔的应用前景。虽然目前GNSS‐声学海底定位技术的研究成果还不能满足海洋时空基准网的建设需求,但其数据处理方法尤其是声速误差精细处理方法,对海洋时空基准网海底部分（海底大地基准）的建设具有重要借鉴意义。介绍了GNSS‐声学海底定位技术的起源,并将其分为静态测量和动态测量两类,同时将声速误差处理方法作为该技术的发展脉络进行梳理,提炼了该技术的3个发展阶段:仅假设海洋声速垂向分层、考虑声速的时域变化、考虑声速的水平梯度。对于仅假设海洋声速垂向分层的阶段,国外学者采用几何结构对称的方式来削弱声速误差的影响;国内学者则主要对声速以外误差源（杆臂矢量误差、时标偏差、姿态角误差等）进行了研究,并用优化随机模型的方式削弱系统误差对定位的影响。对于考虑声速时域变化的阶段,国外学者利用拟合方法（多项式拟合或三次样条拟合）结合参数平滑约束来解算声速的时域变化量,提高定位的稳定性;国内学者基于此细化了参数拟合的方法（考虑参数长周期项的变化特征）,并创新性地提出了水下差分定位算法。对于考虑声速水平梯度的阶段,国内外学者在GNSS‐声学海底定位中解算了声速水平梯度参数,提高了水平方向定位的稳定性,并利用海洋数值模型验证了结果的可靠性。展望了将GNSS‐声学海底定位高精度数据处理方法应用于海底大地基准建设的前景,并引入了小时空尺度声速层析的概念（基于海洋时空基准网的声速误差处理方法）,以期解决数值预报模型不能提供小时空尺度产品的问题,进而为水下潜器提供更高精度的声速误差改正服务。     

通导遥一体化深远海PNT基准及服务网络构想

从海底控制网设计、布设、测量、数据处理和海洋声速场构建几个方面介绍了海底控制网建设的现状。结合国家海洋战略需求和海洋大地测量技术发展,分析了现阶段海底控制网建设面临的问题:认为目前的海底网规模小,缺乏信息共享,功能单一;设计与布设原则未与控制网等级关联,多为定性描述,缺乏操作性;声速场精度和分辨率偏低;测量和数据处理仅考虑了海底控制点的定位问题,难以满足大区域、高精度海洋PNT（positioning, navigation, and timing）基准网建设需求。据此,利用海洋声道的远程通讯和测距定位能力,提出了建设联合北斗/GNSS（global navigation satellite system）定位和通信导航功能的通导遥一体化深远海PNT基准及服务网络的构想,针对该网络的功能和布放设计、高精度、高分辨率海洋声速场模型的构建、各类PNT基准点的测量和整体网平差处理、海底PNT基准点的自校准和自维护、覆盖水域的位置增强服务、目标和环境的遥测和感知服务等几个关键技术问题,提出了实施方法和设想,以期解决当前海洋控制网大区域建设面临的缺乏通讯、布网原则和测量方法不完善、功能单一等问题,为新一代高性能海洋大地测量及PNT导航定位服务网络的建设提供支撑。认为随着长距离布网优化设计、增强位置服务、精确授时和时间同步、远程通讯及网络功能的拓展和应用服务等难题的突破,所构想的网络体系必将引起海洋PNT建设的一次变革。     

Defining a Sea Ice Flag for Envisat Altimetry Mission

This letter presents the development of a sea ice flag algorithm for the Envisat altimetry mission to detect sea ice corrupted sea surface height data within quality control processing. The algorithm takes advantage of having both passive and active microwave sensors on the same platform with coregistered measurements. Its performances have been evaluated based on collocations between the along-track Envisat data with reference maps built from combination of daily grids of sea ice concentration from SSM/I sensors and backscatter cross section from SeaWinds scatterometer on QuikSCAT satellite.     

Assessment of Sentinel-1A data for rice crop classification using random forests and support vector machines

This study developed an approach to map rice-cropping systems in An Giang and Dong Thap provinces, South Vietnam using multi-temporal Sentinel-1A (S1A) data. The data were processed through four steps: (1) data pre-processing, (2) constructing smooth time series VH backscatter data, (3) rice crop classification using random forests (RF) and support vector machines (SVM) and (4) accuracy assessment. The results indicated that the smooth VH backscatter profiles reflected the temporal characteristics of rice-cropping patterns in the study region. The comparisons between the classification results and the ground reference data indicated that the overall accuracy and Kappa coefficient achieved from RF were 86.1% and 0.72, respectively, which were slightly more accurate than SVM (overall accuracy of 83.4% and Kappa coefficient of 0.67). These results were reaffirmed by the government’s rice area statistics with the relative error in area (REA) values of 0.2 and 2.2% for RF and SVM, respectively.