基于机载激光雷达点云的雪道坡度提取算法

滑雪场的雪道坡度信息不仅是雪场规划和建设所需的必要内容,也是滑雪运动员提高成绩、减少运动损伤的重要数据。传统的手动测量方式费时费力,且雪场的低温环境不利于测量人员长时间作业。为了解决滑雪场的雪道坡度计算问题,提出了一种基于激光雷达点云的雪道坡度自动提取算法。使用投影高程差滤波、聚类分割等算法对雪场点云进行预处理,获取雪道边缘特征数据。提出了一种自适应阈值的拟合算法,用于雪道中线的提取,该算法可根据点云数量信息及坐标信息,自动选择阈值拟合雪道中线,消除了实际应用中部分点云数据稀疏或缺失对中线拟合带来的不利影响。数据处理结果表明,该方法计算所得坡度值与多次手动测量平均值之间的相对误差为2.2%。这一方法对激光雷达在雪场目标上的应用、雪场测绘中雪道坡度计算等具有参考意义。     

An improved edge detection algorithm for depth map inpainting

Three-dimensional (3D) measurement technology has been widely used in many scientific and engineering areas. The emergence of Kinect sensor makes 3D measurement much easier. However the depth map captured by Kinect sensor has some invalid regions, especially at object boundaries. These missing regions should be filled firstly. This paper proposes a depth-assisted edge detection algorithm and improves existing depth map inpainting algorithm using extracted edges. In the proposed algorithm, both color image and raw depth data are used to extract initial edges. Then the edges are optimized and are utilized to assist depth map inpainting. Comparative experiments demonstrate that the proposed edge detection algorithm can extract object boundaries and inhibit non-boundary edges caused by textures on object surfaces. The proposed depth inpainting algorithm can predict missing depth values successfully and has better performance than existing algorithm around object boundaries.     

Monte-Carlo calculations of cloud returns for ground-based and space-based LIDARS

A Monte-Carlo model is described which has been developed for calculation of multiply scattered LIDAR returns. Results are shown for the common problem selected by the MUSCLE (MUltiple SCattering LIDAR Experiments) group for intercomparison, which represents a typical ground-based cloud-sensing scenario. This is contrasted with returns from the same cloud sensed by a space-based LIDAR, where multiple-scattering effects are much greater. The magnitude of multiple-scattering effects is seen to be largely determined by the optical depth across the receiver field of view at the cloud.     

基于多小波变换的红外目标探测与识别

针对光电联合变换相关器目标识别的实际应用,对待测红外目标图片进行多小波变换,并利用模极大值法提取其边缘.通过获取更多的轮廓信息,从而提高对目标的识别能力.计算机模拟了常用于红外目标处理的多小波GHM和SA4,实验结果表明:基于GHM多小波提取的边缘能获取大量的图像轮廓信息,其识别结果明显优于SA4多小波.将目标原图的光学相关探测结果与基于GHM多小波提取的边缘图像光学相关探测结果进行比较发现,经多小波预处理后的边缘图像能有效增强相关峰强度.     

Lidar Sensor-Based Object Recognition Using Machine Learning

Journal of Russian Laser Research - Light detection and ranging (LIDAR) sensor provides complicated and large volume of environmental point cloud data that are essential for the target recognition....     

Comparison of indigenously developed micro pulse polarization lidar with EZ lidar profiles

A Micro Pulse Polarization LIDAR (MPPL) has been designed and developed for aerosol and cloud studies at National Physical Laboratory, New Delhi, India (28° 35′ N, 77° 12′E) using a low-energy pico-second pulsed Nd:YAG laser at 532 nm and single PMT detector. This has been used for detecting depolarization characteristics with back-scatter coefficient of atmospheric aerosols and clouds. The back-scattered signals are detected at the emitted wavelength with co-polarization and cross-polarization discrimination with a mirror on stepper motor for aerosols and cloud. Data are obtained by MPPL and are inter-compared with a well-established commercial Leosphere made EZ LIDAR, industry standard at the same site and time, and the results are found to be in good agreement. In the present communication the back-scattered coefficient, aerosols optical depth, depolarization ratio etc. obtained using MPPL & EZ LIDAR are discussed in detail.     

大气云层分布的偏振激光后向散射研究

 利用大气激光后向散射垂直剖面图与消偏振度相结合的方法,分析CALIPSO卫星偏振激光雷达的后向散射信号。以2008年11月23日和24日的CALIPSO卫星数据为例,研究了北京地区大气中云层在可见光和红外光波段的垂直和水平分布特征。利用交互式数据语言IDL(interactive data language)得到大气后向散射强度的垂直分布及其消偏振度,根据垂直剖面图可以直观地观测大气中各成分（如气溶胶、低空云和卷云等）的空间分布情况,并且能清楚地显示大气边界层的高度,由消偏振度可准确获得云层的分布高度及厚度。观测数据的处理结果表明：在海拔高度3 km~7 km存在厚度为2 km~2.5 km的云,其消偏振度约为0.2。     

基于曲面分割的三维点云物体识别

提出了一种基于局部描述符的三维点云物体识别算法.算法首先得到点云的邻域、法线矢量等相关信息,通过邻域进一步得到形状索引值.特征点的提取以形状索引值为依据,以每个特征点为基点对曲面根据欧式距离和矢量夹角分割.分割的曲面片进行等间距划分为多个欧氏距离同心圆,以特征点切平面为基平面投影,并进行等角度抽样,通过抽样点相对特征点的法线矢量及测地距离变化曲线,建立曲面片的二维描述,从而把三维识别转化为二维.根据算法建立模型数据库,给定一个物体,通过和模型数据库中的曲面描述进行比对,得到潜在的识别结果,再通过迭代最近点算法,得到最终的识别结果.最后,通过大量具体实验验证了算法的有效性,并给出了算法的计算复杂度及耗时对比分析,说明了算法的高效性.     

激光雷达辅助驾驶道路参数计算方法研究

针对特定场景辅助驾驶的需求,构建道路场景车载激光雷达三维扫描系统,并设计基于三维激光点云的道路参数提取与计算方法,包括路面宽度、横杆或桥梁或隧道限高、拐弯半径等等,为辅助判断道路的可通过性提供决策依据。测试和对比实验表明,所设计算法效果稳定、精度良好,在效率和自动化程度上有一定优势,可为辅助驾驶提供非常有益的决策参数。     

基于光流估计和自适应背景抑制的弱小目标检测

针对复杂云背景下的弱小目标探测,提出了一种基于光流估计和自适应背景抑制相结合的弱小目标检测算法.首先根据红外图像中云的移动规律,对云背景下的红外图像进行光流分析,提取运动云区.在光流场的计算中结合了云运动的特点以及光流方程的两个约束条件,对传统的基于梯度的光流法予以改进.同时发现移动云区对目标探测的影响较大,为了抑制移动云区对弱小目标的干扰,提出了自适应抑制复杂背景的算法,在光流场分析提取的移动云区中,利用代表背景复杂程度的背景因子,自适应调整分割阈值,抑制复杂背景的干扰.这样只在容易引起虚警的移动云区进行背景抑制处理,简化了计算量,降低了云区对弱小目标的干扰,减少了虚警和误判.实验结果表明该算法可以显著减少云区造成的虚警,并且能够探测出弱小目标.     

Kinetic model for a step edge in epitaxial growth

A kinetic theory is formulated for the velocity of a step edge in epitaxial growth. The formulation involves kinetic, mean-field equations for the density of kinks and "edge adatoms" along the step edge. Equilibrium and kinetic steady states, corresponding to zero and nonzero deposition flux, respectively, are derived for a periodic sequence of step edges. The theoretical results are compared to results from kinetic Monte Carlo (KMC) simulations of a simple solid-on-solid model, and excellent agreement is obtained. This theory provides a starting point for modeling the growth of two-dimensional islands in molecular-beam epitaxy through motion of their boundaries, as an alternative to KMC simulations.     

An adaptive approach for primitive shape extraction from point clouds

In the paper, an adaptive approach for primitive shape extraction from point clouds is presented. The approach extends RANSAC segmentation algorithm in two ways: adaptive primitive shape detection based on histogram analysis of points’ deviations from their corresponding fitted primitive shapes; getting boundary points of the fitted primitive shapes and trimming uniform points of primitive shapes according to boundaries. The data structure of a fitted primitive shape contains two parts: parameter vectors that define the primitive shape; boundary points of the primitive shape. By placing uniform points on primitive shapes and trimming away the points outside of boundaries, we got trimmed uniform points each fitted primitive shapes. Afterwards, we got mesh models of primitive shapes by exerting Delaunay algorithm on each set of trimmed uniform points. At last, we output the final model by calibrating primitive shapes’ positions and orientations and trimming primitive shapes according to their intersection lines.     

Sonic boom effects on circular bridge panels

This paper deals with the response of a curved bridge deck to sonic boom loading. The curved panel is considered as an annular sector plate with the two radial edges simply supported and the circular edges free. The method is equally applicable to other edge conditions. The normal mode method of analysis is used. The response to various durations of boom period as well as different pulse shapes has been determined and presented in the paper.     

星载激光雷达云和气溶胶分类反演算法研究

激光探测对于获取云和气溶胶的垂直廓线,研究大气中云和气溶胶的垂直分布特征以及对全球气候变化的影响意义重大。而星载大气激光雷达云气溶胶分类算法的研究,对于激光雷达数据的参数反演及应用极为重要。针对激光条件下探测的云和气溶胶特有的光学信息和空间分布,结合概率统计与机器学习算法,提出了一种对于云/气溶胶、云相态及气溶胶子类型识别的分类算法,实现了星载激光雷达的大气特征层快速、有效分类。算法采用中国地区2016年CALIOP的观测数据作为样本数据,主要由三部分组成：（1）基于激光探测的云和气溶胶层不同的光学特性以及地理空间分布特征,分别构建了云和气溶胶的γ₅₃₂,χ,δ,Z和lat的五维概率密度函数,以此为基础构建云气溶胶的分类置信函数,并基于此实现了云和气溶胶类型的反演;（2）选取支持向量机(SVM)作为随机朝向冰晶粒子(ROI)和水云分类的算法模型基础,结合云层的γ₅₃₂,χ,δ Z和云顶温度T的概率密度函数构建ROI,水平朝向冰晶粒子(HOI)和水云的分类置信函数以修正SVM误分的特征层以及筛选出水云中少部分的HOI冰云,获得云相态的分类结果;（3）以各气溶胶子类型的光学以及空间分布特性为基础,采用决策树策略的气溶胶子类型识别算法实现了对气溶胶子类型的区分,完成气溶胶子类型的识别。利用现有CALIOP观测结果作为样本数据构建分类数据库,避免了对于地面以及航测数据的依赖,而机器学习则大大简化了算法的实现过程,使得云气溶胶分类更加高效。算法结果与正交极化云气溶胶激光雷达垂直特征层分布数据(CALIPSO VFM)产品对比分析：云层有98.51%一致性,气溶胶有88.43%的一致性,且白天比夜间一致性高。对于云相态分类,可以有效区分出水云和冰云,其中二者水云一致性高达93.44%。在气溶胶子类型反演结果中,可以准确识别出大多数气溶胶特征层子类型。霾、沙尘以及晴空三种典型情况下的反演结果均与CALIOP VFM产品数据具有较好的一致性。其中,霾天的大部分煤烟型以及污染型（污染沙尘以及污染大陆）气溶胶反演结果与VFM具有较好的一致性。沙尘天也能够获得较好的沙尘以及污染沙尘的结果。晴空为数不多的气溶胶层也取得了较为一致的结果。对于实现的星载大气激光雷达特征层分类算法,针对CALIOP激光测量的云气溶胶层的分类进行了重要的改进,在保证一定精度的基础上,简化了算法,提高了数据处理的效率,在下一步工作中,将分别构建不同时段和季节的分类模型以及提高两种不同偏振特性的冰云和气溶胶子类型的分类精度。     

高能闪光照相中Sobel算子的边缘检测方法

 针对高能闪光照相中投影图像的特点,对数字图像处理中具有代表性的Sobel边缘检测算法进行了分析,发现该算法存在检测出的边缘粗和人为进行阈值的选取会造成边缘点的误判的缺点;为此提出了一种阈值确定方法和基于最大梯度的定位准则算法。改进的算法实测结果表明：该方法有效地解决了Sobel算子检测出的边缘过粗的问题,得到的边缘较细,定位精确,提高了边缘检测精度,可用于高能辐射照相中的边缘提取。     

分区拼接网格算法数值模拟超声速复杂流场

以分区拼接网格数值模拟方法为研究对象,对二阶Godunov方法在拼接网格中流场的数值模拟进行了研究,发展了适用于Godunov格式的通量守恒型算法,结合二阶Godunov有限体积法离散非定常Euler方程,数值模拟了捆绑火箭及航天飞机的超声速流场,计算结果正确描述了流场中的激波相交、反射等干扰特性.     

结合机载LiDAR数据与航空可见光影像多层次规则分类建筑物变化检测

建议了一种结合Lidar点云与航空可见光影像的建筑物变化检测新方法,利用多层次规则分类算法解决这两种异元异构数据间建筑物变化检测难题。并建议了一种结合面积阈值的形态学后处理方法,从而形成一套完整的处理流程,可应用于实际生产。最终,利用中国吉林省长春市2010年机载LiDAR点云数据和2009年高分辨率航空影像对该方法的有效性进行了评价,通过与基于支持向量机(SVM)面向对象分类的建筑物变化检测算法比较,进一步对本研究建议的方法进行了验证与分析。结果显示,此方法效果理想,其精度优于基于SVM面向对象分类的建筑物变化检测方法。Kappa系数达到0.90,correctness达到0.87。     

压缩光场重建与深度估计

针对光场深度估计过程中数据量大、边缘处深度估计结果不准确问题,利用压缩感知原理重建光场,提出一种新的多信息融合的光场图像深度估计算法。利用压缩感知重建算法重建5×5视角光场数据,获取光场数据后首先移动子孔径实现重聚焦,然后利用角度像素块散焦线索和匹配线索计算出场景初始深度和置信度。计算图像边缘信息,通过融合初始深度、置信度、边缘信息获取最终深度。实现压缩光场仿真重建,并对仿真光场数据和公开光场数据进行深度估计,实验结果表明:可以仿真重建出5×5视角光场数据,且仿真重建的光场可用于深度估计。该深度估计算法在场景边缘处的深度估计结果边界清晰,层次分明,验证了重建光场深度估计的可行性与准确性。     

一种基于三维激光点云数据的单木树冠投影面积和树冠体积自动提取算法

树冠投影面积和树冠体积是研究单木生物量估测、三维绿量测算等的重要测量因子。针对树冠结构复杂、形态各异,树冠因子难以精确测量等问题,为实现单木树冠投影面积和树冠体积的自动提取和精准测量,以三维激光扫描获取的树木点云数据为数据源,运用平面散乱点集凸包算法——Graham扫描算法和不规则体切片分割累加算法,以VC++6.0和Matlab 7.0混合编程,实现树冠投影面积和树冠体积的自动提取。以实验区选取具有代表性的22株不同树种的样木为研究对象,并与传统人工测量方法进行了对比分析。试验结果表明,树冠投影面积人工测量(A₄)结果与点云数据的自动计算(A_V)结果相关性较好,r=0.964(p<0.01),A₄高于A_V的平均值为25.5%。近似规则几何体方法树冠体积测量(V_C)结果与基于点云数据的自动计算(V_VC)结果之间呈现较强的相关性,r=0.960(p<0.001),V_C低于V_VC的平均值为8.03%。该方法基于高精度单木点云数据,实现了树冠结构的高精度快速重构、单木树冠投影面积和树冠体积的无损自动提取,可以为单木树冠结构的研究提供参考,在精准林业领域具有推广应用价值。     

Photoemission properties of metals

Photoemission has emerged over the last decade as a powerful method for studying electronic band structures. The applicabhty of the method to metals was established by Berglund and Spicer¹ in their landmark experiments on Cu and Ag. It had been believed, largely on the basis of optical absorption data, that the upper edge of the d bands in Cu and Ag lay at 2 and 4 eV below the Fermi level, respectively. Berglund and Spicer's measurements of the photoelectron energy spectra revealed these edges in a spectacular way and also uncovered some of the substructure of the d bands. Since then the technique has undergone some refinement and has been applied to a large number of materials. Photoemission measurements share the advantage of optical absorption measurements in that they are able to probe electronic states well removed from the Ferrni level. They have the added advantage that they give the initial and final energies as well as the frequencies of prominent optical transitions.