Scene 3-D Reconstruction System in Scattering Medium

Research on neural radiance fields for novel view synthesis has experienced explosive growth with the development of new models and extensions.The NeRF(Neural Radiance Fields)algorithm,suitable for underwater scenes or scattering media,is also evolving.Existing underwater 3D reconstruction systems still face challenges such as long training times and low rendering efficiency.This paper proposes an improved underwater 3D reconstruction system to achieve rapid and high-quality 3D reconstruction.First,we enhance underwater videos captured by a monocular camera to correct the image quality degradation caused by the physical properties of the water medium and ensure consistency in enhancement across frames.Then,we perform keyframe selection to optimize resource usage and reduce the impact of dynamic objects on the reconstruction results.After pose estimation using COLMAP,the selected keyframes undergo 3D reconstruction using neural radiance fields(NeRF)based on multi-resolution hash encoding for model construction and rendering.In terms of image enhancement,our method has been optimized in certain scenarios,demonstrating effectiveness in image enhancement and better continuity between consecutive frames of the same data.In terms of 3D reconstruction,our method achieved a peak signal-to-noise ratio(PSNR)of 18.40 dB and a structural similarity(SSIM)of 0.6677,indicating a good balance between operational efficiency and reconstruction quality.

Large-scale spatial data visualization method based on augmented reality

Background A task assigned to space exploration satellites involves detecting the physical environment within a certain space.However,space detection data are complex and abstract.These data are not conducive for researchers'visual perceptions of the evolution and interaction of events in the space environment.Methods A time-series dynamic data sampling method for large-scale space was proposed for sample detection data in space and time,and the corresponding relationships between data location features and other attribute features were established.A tone-mapping method based on statistical histogram equalization was proposed and applied to the final attribute feature data.The visualization process is optimized for rendering by merging materials,reducing the number of patches,and performing other operations.Results The results of sampling,feature extraction,and uniform visualization of the detection data of complex types,long duration spans,and uneven spatial distributions were obtained.The real-time visualization of large-scale spatial structures using augmented reality devices,particularly low-performance devices,was also investigated.Conclusions The proposed visualization system can reconstruct the three-dimensional structure of a large-scale space,express the structure and changes in the spatial environment using augmented reality,and assist in intuitively discovering spatial environmental events and evolutionary rules.

Fast Scene Reconstruction Based on Improved SLAM

Simultaneous location and mapping(SLAM)plays the crucial role in VR/AR application,autonomous robotics navigation,UAV remote control,etc.The traditional SLAM is not good at handle the data acquired by camera with fast movement or severe jittering,and the efficiency need to be improved.The paper proposes an improved SLAM algorithm,which mainly improves the real-time performance of classical SLAM algorithm,applies KDtree for efficient organizing feature points,and accelerates the feature points correspondence building.Moreover,the background map reconstruction thread is optimized,the SLAM parallel computation ability is increased.The color images experiments demonstrate that the improved SLAM algorithm holds better realtime performance than the classical SLAM.

Real-Time Dense Reconstruction of Indoor Scene

Real-time dense reconstruction of indoor scenes is of great research value for the application and development of service robots,augmented reality,cultural relics conservation and other fields.ORB-SLAM2 method is one of the excellent open source algorithms in visual SLAM system,which is often used in indoor scene reconstruction.However,it is time-consuming and can only build sparse scene map by using ORB features to solve camera pose.In view of the shortcomings of ORB-SLAM2 method,this article proposes an improved ORB-SLAM2 solution,which uses a direct method based on light intensity to solve the camera pose.It can greatly reduce the amount of computation,the speed is significantly improved by about 5 times compared with the ORB feature method.A parallel thread of map reconstruction is added with surfel model,and depth map and RGB map are fused to build the dense map.A Realsense D415 sensor is used as RGB-D cameras to obtain the three-dimensional(3D)point clouds of an indoor environments.After calibration and alignment processing,the sensor is applied in the reconstruction experiment of indoor scene with the improved ORB-SLAM2 method.Results show that the improved ORB-SLAM2 algorithm cause a great improvement in processing speed and reconstructing density of scenes.

Scalable point cloud meshing for imagebased large-scale 3D modeling

Image-based 3D modeling is an effective method for reconstructing large-scale scenes,especially city-level scenarios.In the image-based modeling pipeline,obtaining a watertight mesh model from a noisy multi-view stereo point cloud is a key step toward ensuring model quality.However,some state-of-the-art methods rely on the global Delaunay-based optimization formed by all the points and cameras;thus,they encounter scaling problems when dealing with large scenes.To circumvent these limitations,this study proposes a scalable pointcloud meshing approach to aid the reconstruction of city-scale scenes with minimal time consumption and memory usage.Firstly,the entire scene is divided along the x and y axes into several overlapping chunks so that each chunk can satisfy the memory limit.Then,the Delaunay-based optimization is performed to extract meshes for each chunk in parallel.Finally,the local meshes are merged together by resolving local inconsistencies in the overlapping areas between the chunks.We test the proposed method on three city-scale scenes with hundreds of millions of points and thousands of images,and demonstrate its scalability,accuracy,and completeness,compared with the state-of-the-art methods.

One-stage limb Pelnac®reconstruction after removal of skin cancer:safety,efficacy,and aesthetic outcomes

Objective To assesse the outcomes of one-stage limb reconstruction after removal of skin cancers defect.Methods This prospective study was conducted from September 2017 to January 2020 and included 15 patients.All patients underwent extensive tumor resection and one-stage Pelnac®reconstruction of large skin defects,and regular postoperative follow-up was scheduled.At the 6-month follow-up,tumor recurrence and scar quality was assessed using the Vancouver Scar Scale(VSS).None of the patients exhibited infection,wound necrosis,hematoma,seroma,or recurrence.Results All the skin grafts were well accepted by the patients.Nine patients reported normal or near-normal sensory function,while six reported slight sensory loss.No cases of significant functional loss were observed.We enrolled 10 men and 5 women with a mean age of 63.9 years(range:46-78 years).The mean follow-up duration was 20.6 months(range:12-36 months).The skin tumors were located on the feet(n=4),forearms(n=3),and legs(n=8).The malignant tumors included malignant melanomas(13.3%),basal cell carcinomas(33.3%),and squamous cell carcinomas(53.3%).The mean operative time was 40.7 min.Two patients underwent radiotherapy.The average length of hospital stay was 2.6 days.The mean skin defect area was 33.2 cm^(2)(range:16.6-51.6 cm^(2)).The patient satisfaction score(regarding the aesthetic appearance of the grafted area)was 79.7/100,and the VSS score was 3.8.Conclusion Pelnac®dermal templates facilitate efficient and reliable reconstruction of skin defects after skin cancer resection.

动态场景的三维重建研究综述

随着静态场景三维重建算法的不断成熟,动态场景三维重建算法成为近年来的研究热点和研究难点。现有的静态场景三维重建算法对静止的对象有较好的重建效果,一旦场景中对象出现变形或者是相对运动,其重建效果不太理想,因此发展对动态场景的三维重建研究工作是相当重要的。简要介绍三维重建的相关概念及基本知识、静态场景三维重建和动态场景三维重建的研究分类及研究现状;全面总结了动态场景三维重建研究最新进展,将动态场景三维重建按照基于RGB数据源的动态三维重建和基于RGB-D数据源的动态三维重建进行分类,其中RGB数据源下又可划分为基于模板的动态三维重建、基于非刚性运动恢复结构的动态三维重建和RGB数据源下基于学习的动态三维重建,RGB-D数据源下主要总结归纳基于学习的动态三维重建,对各类典型重建算法进行了介绍和对比分析;介绍了动态场景三维重建在医学、智能制造、虚拟现实与增强现实、交通等领域的应用;提出了动态场景三维重建的未来研究方向,并对这个快速发展领域中的各个方向研究进行了展望。

融合威胁情报与知识图谱的网络攻击溯源方法

[研究目的]攻击溯源是网络空间安全保障的重要组成部分,面对网络空间数据海量、异质多元、结构松散等特点,亟需大数据分析与人工智能相结合,有效识别敌手攻击威胁,溯源攻击链和背后的攻击组织,并实施针对性防御。[研究方法]针对攻击威胁特征识别难的问题,提出了知识图谱驱动的网络攻击溯源方法,以脆弱性利用动作为核心构建攻击事件框架,并以事件为单位实施告警关联,重构攻击场景。在此基础上,利用威胁指纹知识图谱,整合已公开的威胁情报知识,并抽取攻击场景中的威胁特征作为指纹,分析两者相似性,溯源攻击者。[研究结论]实验结果表明,该方法能够利用攻击事件框架充实攻击行为的上下文信息,并基于知识图谱有效溯源攻击者,从而利用攻击者已有的威胁情报,增强高级可持续攻击威胁特征识别的全面性。

基于改进FeatDepth的足球运动场景无监督单目图像深度预测

为了在降低成本的同时提高图像深度信息预测的精确度,并将深度估计应用于足球运动场景,提出一种基于改进FeatDepth的足球运动场景无监督单目图像深度预测方法。首先,对原FeatDepth引入注意力机制,使模型更加关注有效的特征信息;其次,将FeatDepth中的PoseNet网络和DepthNet网络分别嵌入GAM全局注意力机制模块,为网络添加额外的上下文信息,在基本不增加计算成本的情况下提升FeatDepth模型深度预测性能;再次,为在低纹理区域和细节上获得更好的深度预测效果,由单视图重构损失与交叉视图重构损失组合而成最终的损失函数。选取KITTI数据集中Person场景较多的部分进行数据集制作并进行仿真实验,结果表明,改进后的FeatDepth模型不仅在精确度上有所提升,且在低纹理区域及细节处拥有更好的深度预测效果。最后,对比模型在足球场景下的推理效果后得出,改进后的模型在低纹理区域(足球、球门等)及细节处(肢体等)有更好的深度预测效果,实现了将基于无监督的单目深度估计模型应用于足球运动场景的目的。

基于KANZI Studio的汽车虚拟仪表三维场景重构技术研究与实现

随着智能座舱技术的快速发展,针对现有汽车仪表场景重构的显示信息不全和效果不达标的问题,二维场景重构技术已无法满足虚拟仪表发展需要。该文利用KANZI Studio汽车HMI专用开发工具,提出了一种实现虚拟仪表三维场景重构效果的方法,并提出效果优化方案。经测试表明,该方法可以解决目前二维场景重构技术的不足,将三维场景重构研究引入到仪表显示端;另将开发周期减少至传统开发工具的50%,有效地降低了三维场景重构的开发难度,能够在保证30帧流畅运行的条件下满足功能要求,最终通过实车验证并在某量产车型上得到应用。

基于高低频特征增强算法的室内场景三维重建仿真

为了提高室内设计效率和质量,提出了一种基于高低频特征增强算法的室内场景三维重建仿真方法。利用小波变换算法提取室内场景单图像高低频特征,通过加权引导滤波算法增强高低频特征图像;采用深度神经网络,预测图像的分割图和深度图,并结合长短时记忆网络估计室内场景单图像平面参数。然后整合分割图、深度图与平面参数,实现三维重建。实验结果表明:该方法可有效提取室内场景单图像高低频特征,增强高低频特征图像和提升图像清晰度;三维重建均方归一化误差最高值为0.15。

灾后社会景气变动趋势:基于后疫情期的考察

灾后社会景气作为表征性因素,是实质性恢复重建工作中长期潜在的隐蔽公理,对于更好促进灾后恢复重建具有重要的社会意义。通过对原有文献中社会景气概念的修复性界定,指出它不仅仅是主观满意度和信心的问题,更涉及客观景象的整体状况。基于后疫情期主客观社会景气的案例,通过实证比较分析进行具体考察,从而揭示疫灾过后经济社会恢复发展和社会景气回升繁荣的艰难之因。在此基础上,提出加强灾后社会景气研究和监测分析,着眼于宏观、中观、微观层面促进灾后恢复重建和社会景气回升。

神经辐射场多视图合成技术综述

如何从图像中渲染出较为真实的虚拟场景一直是计算机图形学与计算机视觉领域的研究目标之一。神经辐射场是一种基于深度神经网络的新兴方法,它通过学习场景中每个点的辐射场来实现较为真实的渲染效果。通过神经辐射场不仅可以生成逼真的图像,而且可以生成具有真实感的三维场景,在虚拟现实、增强现实和计算机游戏等领域有着广泛的应用前景。然而,其基础模型存在训练效率低、泛化能力差、可解释性不足、易受光照和材质变化影响以及无法处理动态场景等问题,在某些情况下无法获得最佳的渲染结果。大量基于此研究的工作陆续展开,且在效率和精度等方面都取得了出色的成果。为了跟踪该领域最新研究成果,对近年来神经辐射场领域的关键算法进行回顾和综述。首先介绍了神经辐射场的产生背景及原理,对后续关键改进模型进行分类探讨。主要涵盖以下几个方面:对神经辐射场基本模型参数的优化,在渲染速度与推理能力方面的提升,对空间表达和光照能力的改善,针对静态场景相机位姿估计与稀疏视图合成方法的改进,以及在动态场景建模领域的发展。对各种模型的速度与性能进行分类对比与分析,并简要介绍了该领域主要模型评估指标与公开数据集。最后对神经辐射场未来发展趋势进行展望。

基于深度学习无人机影像道路实景三维修复

针对基于无人机倾斜影像实景三维重建中,移动目标对道路实景三维重建造成几何变形和纹理失真的问题,提出一种基于深度学习的无人机影像道路实景三维修复方法。首先,通过添加注意力机制的深度学习网络模型YOLOv8对影像中目标进行检测;其次,在得到影像对应目标标记的基础上,根据已生成三维Mesh模型中各三角面在可视影像集合中的投影位置,结合影像所标记目标的对应范围,统计各三角面的类别信息以此判定移动目标;最后,利用移动目标判定结果对移动目标造成的三维模型几何变形及纹理错误进行修复,实现道路实景三维重建。结果表明:改进的网络模型较YOLOv4、YOLOv5和YOLOv8模型,平均精度(mAP)值平均提升10.82%,移动目标判定准确率达97.43%。与流行国外商业软件相比,所提方法重建修复效果更佳,自动化程度更高。

一种基于对偶学习的场景分割模型

城市场景分割等复杂任务存在特征图空间信息利用率低下、分割边界不够精准以及网络参数量过大的问题。为解决这些问题,提出了一种基于对偶学习的场景分割模型DualSeg。首先,采用深度可分离卷积使模型参数量显著减少;其次,融合空洞金字塔池化与双重注意力机制模块获取准确的上下文信息;最后,利用对偶学习构建闭环反馈网络,通过对偶关系约束映射空间,同时训练“图像场景分割”和“对偶图像重建”两个任务,辅助场景分割模型的训练,帮助模型更好地感知类别边界、提高识别能力。实验结果表明,在自然场景分割数据集PASCAL VOC中,基于Xception骨架网络的DualSeg模型的mIoU和全局准确率分别达到81.3%和95.1%,在CityScapes数据集上mIoU达到77.4%,并且模型参数量减少18.45%,验证了模型的有效性。后续将探索更有效的注意力机制,进一步提高分割精度。

多视影像三维线云重建技术及其智能化发展展望

三维线云是具有几何结构与语义信息的三维线段集合,可以作为实景三维构建中高效的引导,控制与抽象表达元素,弥补点云的边缘描述缺陷与初始结构的缺失,是改变传统三维模型“一张皮”(不同对象相互粘连,难以指导空间分析与决策)的重要结构化特征。然而,如何从多视影像中重建好用的线云,将线云在实景三维中用好,一直是本领域的难点问题。本文回顾三维线云的发展,介绍相关的重建方法,分析存在的难点与不足;结合从数字化走向智能化的测绘科技转型背景,论述线云在实景三维中建什么,如何建与怎么用,对线云重建和应用的智能化发展予以介绍和展望,希望为实景三维重建与线云研究的相关人员提供参考。

城市建筑三维实体自动化建模技术集成与实践

城市建筑实体是实景三维中国建设中基础地理实体生产的重要内容之一。针对建筑物体块三维模型,现有基于点云数据的建模算法和软件普遍存在自动化程度低、精细程度不足、规模化生产受限等问题,本文提出了面向城市建筑实体的自动化三维重建方法。首先,采用点云分割实例算法识别独立建筑物点云,自动提取并优化其底面轮廓信息;然后,将点云数据进行精简结构管理,通过欧式聚类获取各个顶面构件点集并映射拟合为多边形,引入底面轮廓作为硬约束条件进行后处理精化与合并;最后,研发实现了城市建筑实体自动化三维重建软件,该软件集成了大规模建模、编辑、可视化及质检等功能。上述关键技术及方法已在宁波、南昌和楚雄3个实景三维建设项目进行了验证。验证结果表明,本文方法支持万级以上建筑物体块三维模型快速重建,且建模速度优于16栋/s,可为实景三维建筑实体的规模化生产建设提供有力支撑。

融合场景深度估计和视觉传达的复杂光照图像虚拟重建

复杂光照图像虚拟中受到光照强度不均衡性影响导致重建效果不好,为了提高复杂光照图像虚拟重建效果,提出基于融合场景深度估计和视觉传达的复杂光照图像虚拟重建方法。针对不同场景深度混频光照的相互干扰采用相关匹配降噪方法实现图像降噪处理,以光照图像低亮度区域内亮度值中位数作为场景深度的参考值,采用全局特性和局部细节特征拟合的方法实现对复杂光照图像的场景深度检测和视觉跟踪拟合,采用HSV空间特征分解方法实现对不同场景中光照图片亮度通道融合处理,提取场景物体边缘、纹理等细节信息,根据场景深度检测和全局对比度融合下的视觉传达效果实现复杂光照图像虚拟重建。测试结果得知,采用该方法进行复杂光照图像虚拟重建的视觉表达能力较好,重建后的图像细节展示能力较强,能准确重建暗区域内隐藏的图像信息,两个数据集图像的峰值信噪比较高,均方根误差较低,分别为45.63 dB、53.21 dB和0.366、0.265,且重建时长短,仅为1.5 s,具有较强的重建性能。

基于数字孪生的智能掘进工作面三维场景再现技术研究

目前井下掘进设备作业一般采取视距遥控或远程视频监控方式,巷道断面成形质量取决于人工操作经验或视频数据传输质量。针对该问题,结合智能掘进系统实时监控和真实场景再现需求,研究了基于数字孪生的智能掘进工作面三维场景再现技术,完成了数字孪生系统架构设计、传感器数据采集处理、数字孪生模型创建、模型驱动算法开发、显示界面设计。根据神东某矿掘锚一体机工作面的工况条件,开发了一套三维场景再现系统,内嵌数字孪生体、数据监控与驱动仿真模块,集成设备传感器、动作指令等数据,实现对掘进系统的物理设备状态实时监测、数字三维模型同步展示、数据分析管理以及智能掘进工艺仿真。实践证明,该系统能够实时同步精准映射设备状态、巷道空间信息,实现场景真实再现。与视频远程控制的方式相比,改善了操作体验和准确度,有效提升了掘进智能化水平,保障了巷道掘进效率和施工质量。

大规模场景运动恢复结构研究综述

运动恢复结构旨在基于图像间的局部特征匹配解算相机在全局统一坐标系下的绝对位姿,是基于图像的三维重建中的关键问题.近年来,随着采集设备、计算资源以及理论方法的发展,运动恢复结构研究已逐渐由实验室小规模可控场景向室内外大规模实际场景扩展,并取得了成果显著的理论方法与实际应用.文中从实际应用出发,综述了面向大规模场景三维重建的运动恢复结构研究领域的最新成果;专注于运动恢复结构中相机位姿解算核心问题,并全面介绍了其中的最新成果按照解析式与学习式方法进行;在此基础上,为助力社区发展,讨论与分析的运动恢复结构当前研究进展与未来发展态势.