非量测相机制约迭代检校法及水下测量应用

传统的非量测相机检校方法大多依赖于高精度的室内或室外控制场,并对检校环境有较严格的要求。故提出了一种完全抛开控制点的新的相机检校方法,即线面角制约的相机迭代检校法。将该方法分别应用于陆地和水下环境,并与有控制点的相机迭代检校法进行对比分析,试验结果表明新方法的检校精度均小于1个像素,在陆地环境下,新方法的检校精度优于有控制点的相机迭代检校法。同时利用新方法进行了水下三维重建和量测,量测精度达到亚毫米级。试验结果表明新方法具有较好的灵活性和适应性,在多个环境下具有实际应用的价值。     

IMU视轴偏心角的两步法检校

建立了IMU视轴偏心角模型,重点研究了两步法偏心角检校中基于欧拉角、四元数和微分旋转矩阵的偏心角解算方法,并推导了欧拉角和单位四元数两种模型的线性化公式。实验结果表明,欧拉角模型具有最高的解算精度,四元数模型具有最强的适应性,二者均适用于IMU视轴偏心角的检校。微分矩阵模型仅在偏心角为小角度时有效,是一种近似方法,应谨慎使用。     

机载LIDAR系统校准方案优化设计

为优化机载LIDAR(Light Detection and Ranging)系统校准方案,基于天津、烟台和舟山3个标准检校场的试验数据,分析比较不同检校场地、不同飞行航高和不同飞行航速对激光扫描数据检校结果的影响.探索激光扫描器脉冲能量大小、激光扫描仪视场角、扫描频率等参数设置对激光扫描数据检校结果的影响,最终提供了...     

基于自适应形态学的甲藻显微图像顶刺提取

为有效地提取甲藻生物形态学特征,提出一种自动分割藻种顶刺细节特征的形态学方法,该方法首先将目标旋转至其主轴方向,引入了像素宽度的概念,以藻种目标像素宽度直方图和面积分布为判别依据,自动判定最佳结构元尺寸,最后联合滤波、区域归并等形态学操作提取顶刺区域。实验证明,该算法针对不同甲藻细胞目标可自动计算最佳结构元素大小,精确提取顶刺区域,具有精度高、速度快的优点。     

地面三维激光扫描仪自检校方法研究

在深入分析地面三维激光扫描仪系统误差的基础上,研究了其系统误差检校主要方法.提出了扫描仪系统误差自检校方法,设计了基于控制点的室内自检校场,通过对GLS- 1500扫描仪的自检校实验,验证了自检校方法的可行性,并通过显著性检验,为优化扫描仪系统误差参数结构提供了依据.     

基于三维DLT方法的非量测数码相机检校

低空无人机遥感系统应用日益广泛,其一般搭载传感器为非量测型数码相机,为了保证测量精度必须进行相机检校,以确定其内方位元素和畸变差系数。采用室外三维控制场的DLT(直接线性变换)方法,编制了基于三维DLT方法的非量测型相机的检校软件,该软件具备相机检校参数解算和物方空间坐标解算功能。对于室外三维控制场的测量数据和一海岛的无人机航摄数据,分别使用和不使用相机检校参数进行处理,试验结果显示使用检校参数可以获得更好地测量精度,表明了室外三维检校场方案和编制的相机检校软件是正确可行的,可实现非量测数码相机的检校。     

海洋重力测量误差补偿简化方法研究

针对实际应用的需要,本文对作者先前提出的两种海洋重力测量误差补偿方法进行了简化,把原来严密的自检校平差简化为两步处理法,深入分析了简化方法的技术特点和适用范围。最后使用一个实际观测网数据验证了两步简化方法的有效性和可靠性。     

保持局域特征的多源海冰图像融合与分类

本文提出一种保持局域特征的多源海冰图像融合方法，并在此基础上进行海冰分类。本文提出的多源海冰图像融合方法包括保持空间局域融合和保持特征局域融合两方面。首先，通过学习得到投影矩阵和相似矩阵。投影矩阵将多源像素进行投影变换，得到保留像素空间局域特性的融合向量。相似矩阵度量像素特征间的相似性，通过拉普拉斯特征分解，得到保留像素特征局域相似性的融合向量。然后，将空间融合向量和特征融合向量进行像素综合，得到融合图像。在此基础上，本文设计一种滑动集成分类方法进行融合图像像素分类。提出的分类方法利用滑动集成的特点，在分类时增强刻画了海冰局域特性。由于本文的保持局域融合框架不仅刻画了海冰在物理空间中的邻接关系，而且考虑不同海冰类型的特征关系，因此其在多源图像（多光谱和合成孔径雷达（SAR）图像）的海冰分类任务中表现优异。实验结果表明本文提出的基于保持局域特征融合的多源海冰图像分类方法有效提升了海冰分类精度。     

基于多尺度小波特征的高光谱影像亚像素目标识别

提出了一种基于多分辨率小波高频特征系数的高光谱遥感影像亚像素目标识别方法。首先利用多尺度小波变换将光谱信号分解为不同尺度的高频特征信号,然后借助接收操作特性曲线(ROC)和马氏距离投影寻踪求取一维最佳识别特征,最后通过高斯最大似然决策函数求解亚像素目标的存在概率。通过38种小波函数的高光谱数据实验证明,该方法对亚像素目标的识别效果较好。     

同步气象卫星视地理坐标与像素直角坐标的转换关系及误差分析

利用WGS-84坐标系模型及纬度的定义办法,根据同步气象卫星的视成像原理,导出了同步气象卫星云图像素的视地理坐标与直角坐标的转换关系,建立起一种像素地理坐标定位的理论模型,并对模型进行了误差分析,用实收云图验证表明,推理正确,结论可靠,为同步气象卫星数字化云图资料的再定位和下游延伸研究和应用提供了一种可靠的定位方法.     

基于单位四元数的小像幅面阵CCD相机检校

介绍了面阵CCD相机的径向畸变、偏心畸变模型,建立了用于非量测相机标定的扩展共线条件方程.引入了四元数姿态描述方法,利用单位四元数替代外方位角元素(ω,ψ,к)构建旋转矩阵,建立了基于单位四元数的相机标定数学模型.采用室内三维控制场对小像幅面阵CCD相机进行了标定实验,取得了优于0.2个像素的定位精度.     

水下视觉SLAM相机成像畸变纠正研究

针对水下视觉SLAM相机成像畸变问题,分析了水下机器人相机成像畸变原因,构建了纠正模型计算畸变系数以获取正确的像素坐标,实验证明在空气和水两种不同介质中,相机内参和畸变系数均有较大差异,而在不同水体中,仅畸变系数存在变化。表明在进行水下SLAM时,采用当前实验水体标定参数可以获取更好的定位和地图构建结果。     

A quantitative striping removal algorithm for HY-1 CCD data

Striping is a common phenomenon in remote sensing imagery acquired by most spaceborneand airborne multispectral sensors. In contrast to the destriping of the images of optoelectronic mechanical scanners, fewer work has been done for push-broom CCD images. Based on the characteristics of HY -1 push-broom multispectral CCD camera and its prelaunch radiometric calibration results, the striping features are analyzed and a quantitative striping removal algorithm is proposed. The algorithm is to obtain the inter-detector equalization curves to reflect the differences between detectors by statistic analysis of the satellite imagery data set, and image stripes arising from the nonuniformity of the detectors of the CCD array can be removed using the estimated equalization curves. The preliminary results show that this method can effectively remove the stripes and preserve the radiometric accuracy of the raw data at the same time.     

Using a digital camera for water quality measurements in Galway Bay

A method for using the red–green–blue (RGB) digital output from a digital camera, a Nikon Coolpix 885, to estimate water quality parameters is described. To eliminate reflection from the water surface, a tube is fitted around the camera lens which breaks the surface such that only the upwelling light from below is measured. To obtain consistency between stations, RGB values used for analysis are taken where the total cameras response (R + G + B) is a constant value, derived from pictures taken at a range of exposures.The method has been successfully used in Galway Bay, where the optical characteristics of the coastal water are dominated by yellow substance (or coloured dissolved organic material – CDOM) and chlorophyll. A strong linear relationship was found between CDOM and the Red/Blue camera output. The form of the relationship was consistent with previous measurements using more conventional narrow band irradiance sensors. In addition, a log–log relationship between chlorophyll and the Green/Blue camera output was also found. Some problems exist with the use of the camera, particularly with the cameras automatic functions such as ‘white balance’ which may artificially change the RGB output. Nevertheless, the use of a camera for simple water quality measurements has been demonstrated.     

卫星传感器的地面几何检校模型分析

针对高分辨率卫星 CCD 线阵传感器,分析了像元尺寸变化模型、CCD 在焦平面内旋转变化模型、传感器镜头光学畸变模型;基于 ALOS 卫星的内方位附加参数模型,构建了卫星传感器内方位自检校综合模型;分析了卫星传感器严格成像模型及外方位检校参数内容,给出了自检校光束法区域网平差解算方案;结合外方位元素 PPM 内插模型,利用 SPOT5 模拟数据进行区域网平差实验,实验结果验证了模型方案的可行性。     

Quantitative photogrammetric analysis of digital underwater videoimagery

This paper presents a photogrammetric model for digital underwater video imagery, which has been mostly applied to qualitative analysis in the marine environment. With this model, quantitative analysis of underwater images is possible, e.g., to locate positions, calculate sizes, and measure shapes of objects from image features. The underwater photogrammetric model is based on a three-dimensional optical ray tracing technique which rigorously models imaging systems with multilens configurations and multiple refractions. The calibration procedure with two independent phases has been proven to be efficient in simplifying the computation and improving the calibration accuracy. With the current imaging system configuration and photogrammetric model, an accuracy of 0.8 cm in lateral directions and 1.2 cm along the depth direction for objects located about 2-3 m from the camera system in the object space is attainable. A PC-based digital underwater photogrammetric prototype system has been developed to implement the underwater photogrammetric model     

Analysis and evaluation of coral reef integrity based on borehole camera technology

At present, very few drilling surveys have been conducted on coral reefs, and comprehensive geological info on coral reefs is difficult to obtain. However, research on coral reefs is of great importance to development and exploitation of maritime resources. In the survey of Chenhang Island in Paracel (Xisha) Islands, we applied the digital borehole camera technology the first time in China, which resulted in a large amount of first-hand geological data. We utilized the rock mass integrity index (RMDI) method in our analysis of coral reef integrity, and the value of coefficient α in the RMDI was determined to evaluate the size effect of cavities on coral reefs. We then developed the correlation between RMDI values and coral reef integrity with references to the rock mass classification standard, which enabled the engineering evaluation of coral reef integrity. By statistical analysis of camera data gathered along the CK2 borehole for approximately 1-km length, we performed an integrity evaluation of the entire borehole length using the RMDI classification and created a coral reef stratum development diagram that was consistent with the actual cores. The results indicated: (1) the RMDI is a feasible approach for the analysis of coral reef integrity, (2) the classification system based on RMDI is feasible in the integrity evaluation of coral reefs, (3) the values of coral reef integrity obtained using the RMDI method are accurate, and (4) borehole camera survey and statistical analysis are important in the study of coral reefs.     

基于深度学习的视频观测潮位技术研究——以厦门高崎码头为例

近岸潮位观测是海洋工程应用、海岸防灾减灾、海岸带管理以及海洋有关科研工作中最基础的工作之一。文章基于视频图像深度学习的方法,使用YOLOv5目标检测算法从安装在近岸的固定摄像机拍摄的视频帧中提取潮汐水位特征进行潮位分析。研究采用厦门高崎码头的分辨率为1920×1080的高清摄像头2023年2月的影像数据作为训练集和验证集,2023年3月的影像数据作为测试集,利用岸边验潮井逐时潮位数据进行标注,采用YOLOv5目标检测算法来训练。计算结果显示,通过视频观测潮位在训练集和测试集上的误差分别为3.9 cm和5.3 cm。视频中1个像素点代表3.8 cm,因此潮位观测的平均误差为像素级。研究表明在近岸通过高清摄像头基于图像深度学习进行潮位观测的方法是可行的,观测精度取决于图像目标物的分辨率。     

Estimation of surface chlorophyll on an exposed mudflat using digital colour-infrared (CIR) photography

The usefulness of field-based digital Colour-InfraRed (CIR) photography to quantify concentrations of chlorophyll on the surface of exposed mudflats is investigated. Multiple images, each 626 mm by 467 mm, were acquired during Austral summertime using a Duncantech three-band CIR camera from two areas of mudflat in the upper reaches of Sydney Harbour. Sediment samples were obtained from within the field of view of the camera and their chlorophyll concentration was estimated spectrophotometrically. After the camera images were normalised to compensate for the effects of variations in the intensity of downwelling solar radiation, chlorophyll was estimated for each 0.9 mm square pixel using a suite of five different vegetation indices. Regression analysis was used to determine the strength of the relationship between the index values and the estimates of chlorophyll from the in situ samples. Indices constructed from near-infrared and red bands were found to have the strongest relationships with in situ chlorophyll estimates (R² ranging from 0.28 to 0.79) and indices derived from near-infrared and green bands the weakest (R² ranging from 0.16 to 0.22). The vegetation indices highlighted complex small-scale variability in chlorophyll distribution that was not evident in the original camera images. These findings indicate that field-based CIR photography will provide a useful tool for the non-destructive determination of benthic chlorophyll.