三维激光成像系统的点云模型真彩色处理方法

三维激光成像系统获得的点云数据精度高并且保有高精度的高程信息,然而特征信息不明显,没有保留全部的点和点的拓扑关系,无法直接判读地面物体;影像数据空间分辨率和精度低,并且缺少高程信息,然而其影像的特征连续,光谱信息丰富.因此,进行点云数据与影像数据融合方法的研究是目前三维激光成像系统研制的前沿性课题.针对点云数据与影像数据融合方法的研究,提出了一种三维激光成像系统点云模型真彩色处理的方法.实验结果显示,采用所提出的方法,不仅能够获得被扫描物体的位置信息和大小信息,而且能够得到被扫描物体的表面纹理细节信息.     

Color Image Classification Using Support Vector Machines

0IntroductionThefastadvancementofdigitalimagingtechnologyhasresultedintheexponentialin-creaseofimagedatainbothprofessionalarchivesandpersonalleisurecollections.Effectivelymanag-inglargeimagecollectionsandmakingthemeas-ilyaccessibleposessignificanttechnica…     

TopPixelLoss:类别不均衡的遥感影像语义分割损失函数

针对遥感影像中类别不均衡的小目标分割效果不理想的问题，提出了一种类别不均衡小目标二分类分割的损失函数——TopPixelLoss损失函数。首先计算出每个像素的交叉熵，然后将所有像素的交叉熵按从大到小进行排序，随后确定一个K值作为阈值，筛选出前K个交叉熵最大的像素，最后对于筛选出的K个像素交叉熵取平均，做为损失值。在ISPRS 提供的 Vaihingen 数据集上，使用PSPNet网络与普通交叉熵、FocalLoss、TopPixelLoss三种损失函数分别对车辆进行二分类分割试验。结果表明，不同的K值，使用TopPixelLoss损失函数的平均交并比（MIoU）、F1-score、准确度（ACC）都最高;当K值为5×104时效果最佳，MIoU、F1-score、ACC分别比FocalLoss提高了3.0%、5.0%、0.1%。TopPixelLoss损失函数是一种针对类别不均衡分割非常有效的损失函数     

Visual inspection of aircraft skin: Automated pixel-level defect detection by instance segmentation

Skin defect inspection is one of the most significant tasks in the conventional process of aircraft inspection. This paper proposes a vision-based method of pixel-level defect detection, which is based on the Mask Scoring R-CNN. First, an attention mechanism and a feature fusion module are introduced, to improve feature representation. Second, a new classifier head—consisting of four convolutional layers and a fully connected layer—is proposed, to reduce the influence of information around the area of the defect. Third, to evaluate the proposed method, a dataset of aircraft skin defects was constructed, containing 276 images with a resolution of 960 × 720 pixels. Experimental results show that the proposed classifier head improves the detection and segmentation accuracy, for aircraft skin defect inspection, more effectively than the attention mechanism and feature fusion module. Compared with the Mask R-CNN and Mask Scoring R-CNN, the proposed method increased the segmentation precision by approximately 21% and 19.59%, respectively. These results demonstrate that the proposed method performs favorably against the other two methods of pixel-level aircraft skin defect detection.     

飞机复杂蒙皮拉形数值模拟系统开发及关键技术研究

针对飞机复杂蒙皮拉伸成形特点,基于静态隐式算法开发了复杂蒙皮拉形数值模拟系统STRETCH FORM。系统分为前置处理、核心求解和后置处理3个模块。介绍了系统各模块的主要功能以及机构运动、静态隐式有限元算法和接触搜索算法等关键技术理论。利用商业软件MARC和PAM STAMP进行了S形蒙皮拉形的模拟计算对比,并对某型号机的进气道蒙皮零件进行了生产验证,结果表明系统具有较高的可靠性和计算精度。     

变背景下斑点状散布目标自动识别算法

从自动目标识别原理出发,通过对变背景下斑点状散布目标特性的分析,发现了该类目标图像中背景、噪声和斑点状散布目标的固有规律。进而,从图像像素灰度这一最基本的要素着手,研究了如何在大量的序列图像中搜索目标帧位置的方法。然后,基于目标轮廓提取和串行边界分割技术,解决了变背景下斑点状散布目标自动识别的难题。实验结果证明:该算法具有高效、快速、稳定、适用等优点,并为相关领域内的应用提供了宝贵的参考价值。     

空间目标快速轮廓特征提取与跟踪技术

为满足空间目标交会对接任务中高精度、快速的测量要求,提出了一种空间目标快速轮廓特征提取与跟踪技术。该算法首先从初始帧图像中分割定位目标所在局部区域,作为目标连续跟踪的初始值;其次基于初始帧目标局部区域完成对初始帧目标边缘特征的检测及细化处理;最后采用Hough变换完成对初始帧目标边缘的检测及细化后的局部图像轮廓直线的提取,分别选取目标轮廓四方向最优的直线参数作为最终目标轮廓直线获取的效果,并采用梯度最大法则实现两两求交获取的轮廓特征的优化提取。在目标逼近过程中,结合相邻帧图像间目标尺度动态变化的关联性,根据初始帧提取目标轮廓特征的先验信息,确定目标在第二帧图像中的轮廓位置,并依次根据上一帧图像的轮廓位置信息定位目标在当前帧所在的区域,通过局部处理实现序列图像轮廓区域特征的连续跟踪。该算法无需遍历整个图像,所需处理的目标区域大幅减小,能够有效克服由目标图像较多边缘干扰导致的轮廓提取效果差及处理速度慢的缺点,具有速度快、准确性强、稳定性高等优点。     

飞机蒙皮镜像铣加工稳定性分析

镜像铣技术是近年来提出的一种针对大尺寸薄壁件的加工方法,其加工环保、高效,具有逐步取代传统化铣加工的趋势。针对飞机蒙皮镜像铣加工过程中的颤振问题,首先,根据镜像铣加工特点,建立工艺系统颤振稳定性极限的预测模型;其次,通过有限元方法对蒙皮工件进行模态分析,分析蒙皮不同加工位置动力学特性变化导致的铣削稳定性变化,并通过实验测量获取工件动力学参数,对不同加工位置的稳定性做出预测;最后,开展加工实验,基于非接触测量方法,在线监测镜像铣加工区域的振动位移,通过时、频域信号对加工状态进行辨识,揭示了镜像铣加工过程的稳定性变化及失稳机制,验证了所提方法的准确性,在工程领域具有较高的实用价值。     

简析PW1100G进气道唇口蒙皮损伤修理

以PW1100G发动机进气道为例,简析其主要结构特点,总结整理典型唇口蒙皮损伤的修理方法,并结合实际维修过程中的经验数据,对不同程度损伤修理方法的选用进行梳理。     

Satellite-observed biological variability in the equatorial Pacific during the 2009–2011 ENSO cycle

The event of 2009–2011 El Niño Southern Oscillation (ENSO) provides an opportunity to gain insight into the biological variability of the equatorial Pacific Ocean for an entire ENSO cycle with satellite and in situ observations. Even though El Niño and La Niña in general led to respectively weakened and enhanced chlorophyll-a concentration and net primary production (NPP) along the equatorial Pacific Ocean during the 2009–2011 ENSO cycle, biological responses were highly disparate along the equator and attributed to different driving mechanisms. In the eastern equatorial Pacific east of 150°E, the El Niño-La Niña biological change was in general small except for the transition period even though sea surface temperature (SST) showed over ∼5 °C drop from El Niño to La Niña. In the central-eastern (170°W–140°W) equatorial Pacific, moderate change of biological activity is attributed to the changes of thermocline driven by the eastward propagating equatorial Kelvin waves and changes of zonal currents and undercurrents. Highest biological response in this ENSO cycle was located in the central (170°E–170°W) and central-western (150°E–170°E) equatorial Pacific with quadruple chlorophyll-a concentration and over ∼400 mg C m⁻² d⁻¹ increase of NPP from El Niño in 2009 to La Niña in 2010. However, spatial pattern of ENSO biological variability as represented with NPP is not exactly the same as chlorophyll-a variability. Wind-driving mixing of nutrients and eastward advection of the oligotrophic warm pool waters are attributed to this significant biological variability in this region.