位姿估计
文章平均质量分 70
冼yy
这个作者很懒,什么都没留下…
展开
-
Stereo Plane SLAM Based on Intersecting Lines
Stereo Plane SLAM Based on Intersecting LinesAbstractIntroductionRelated WorkPlane features from interscting linesSLAM system based on computed planesEvaluationConclusionsAbstract文中提出一种在双目相机的面类型的视觉SLAM方案,通过俩个线性特征信息来构建一个plane元素信息,通过获取3D空间内的端点和线性特征信息的方向向量来原创 2020-10-09 18:18:57 · 364 阅读 · 0 评论 -
GMS: Grid-based Motion Statistics forFast, Ultra-robust Feature Correspondence
GMS: Grid-based Motion Statistics forFast, Ultra-robust Feature CorrespondenceAbstractIntroductionOur approachGrid framework for fast scoringExperimentsConclusionAbstract文中提出一种新型的特征点匹配的方案,GMS,通过考虑运动的平滑性来提升特征信息的匹配质量,通过将匹配数量转换为匹配质量。Introduction文中主要解决的问题原创 2020-09-29 21:43:36 · 432 阅读 · 0 评论 -
Hand-held Monocular SLAM Based on Line Segments
Hand-held Monocular SLAM Based on Line SegmentsAbstractIntroductionImage LineSpatil LineMethorodlogy of Monocular SLAMExperimental ResultsConclusionAbstract本文提出了一种基于线性特征信息的视觉SLAM方案,通过普吕克坐标表示直线,通过NLT算法对直线进行匹配,之后通过卡尔曼滤波算法对直线的位置信息进行校准,最后通过构建精准的地图信息。Introdu原创 2020-09-28 21:48:41 · 252 阅读 · 0 评论 -
PLS-VIO: Stereo Vision-inertial Odometry Based on Point and Line Features
PLS-VIO: Stereo Vision-inertial Odometry Based on Point and Line FeaturesAbstractIntroductionRelated WorkSystem OverviewRepresentation and Detection of line featureCoupled Optimization Model for Stereo Vision Inertial NavigationAbstract文中提出了一种结合点和线特征的视觉原创 2020-09-28 09:36:09 · 741 阅读 · 0 评论 -
Realtime edge-based visual odometry for a monocular camera
Realtime edge-based visual odometry for a monocular cameraAbstractIntroductionAbstract文中通过结合特征法和直接法俩种主流视觉SLAM来实现更加精准的SLAM方案,首先,通过特征法Introduction原创 2020-09-27 16:41:36 · 374 阅读 · 0 评论 -
FastORB-SLAM: Fast ORB-SLAM method with Coarse-to-Fine Descriptor Independent Keypoint Matching
FastORB-SLAM: Fast ORB-SLAM method with Coarse-to-Fine Descriptor Independent Keypoint MatchingAbstractIntroductionRelated WorkSystem OveriveCoarse-to-Fine descriptor-independent krypoint matching methodExperimentsConclusionAbstract相对于ORB_SLAM而言,本文所提出的方案原创 2020-09-25 15:29:40 · 777 阅读 · 0 评论 -
ORB: an efficient alternative to SIFT or SURF简介
ORB: an efficient alternative to SIFT or SURFAbstrauactIntroductionRelated WorkoFAST:FAST Keypoint OrientationrBRIEF:Rotation-Aware BriefScalable Matching of Binary FeaturesApplicationsConclusion补充Abstrauact特征信息匹配问题在计算机视觉中的应用十分广泛,其中针对特征点信息的匹配已经有很多的方案,但在运原创 2020-09-23 22:00:31 · 397 阅读 · 0 评论 -
Speeded-Up Robust Features (SURF)笔记
@[TOC](Speeded-Up Robust Features (SURF))AbstractIntroductionRelated Work补充:【特征匹配】SURF原理与源码解析(一)SURF算法surf算法原理-包你明白surf过程【CV学习5】SURF算法详解原创 2020-09-23 12:17:22 · 428 阅读 · 0 评论 -
Distinctive Image Features from Scale-Invariant Keypoints
Distinctive Image Features from Scale-Invariant Keypoints链接很不错的博文一篇比较全面的博文原创 2020-09-21 21:26:09 · 648 阅读 · 0 评论 -
RGB-D Visual Odometry Algorithm Based on Improved ORB for Indoor Environments
RGB-D Visual Odometry Algorithm Based on Improved ORB for Indoor EnvironmentsAbstractIntroductionSystem IntroductionExperimental Results and AnalysisConclusion and Future workAcknowledgementAbstract本文提出一种结合四叉树进行特征信息提取的算法以提高位姿估计的精度。Introduction文中提出的一种通过原创 2020-09-13 19:09:20 · 144 阅读 · 1 评论 -
REAL-TIME MONOCULAR VISUAL SLAM BY COMBINING POINTS AND LINES
REAL-TIME MONOCULAR VISUAL SLAM BY COMBINING POINTS AND LINESAbstractIntroductionSystem OverviewLine Feature in Visuak SLAMCombined Treatment of Points and LinesExperiment al ResultsConclusionAbstract本文提出一种基于特征点和特征线的一种视觉里方法,提出了一种对特征点和线性特征信息进行并行处理的方式,线性特征原创 2020-09-12 19:39:16 · 220 阅读 · 0 评论 -
Real-Time RGB-D SLAM with Points and Lines
Real-Time RGB-D SLAM with Points and LinesAbstractIntroductionSystem OverviewMethods in Tracking ProcedureExperimentsConclusionsAbstract一个实时的基于特征点和线性特征信息的RGBD-SLAM方案,在特征点提取部分通过一定的改进在一等程度上避免了光度变化对定位效果的影响。Introduction介绍了视觉SLAM的发展,其中基于特征点类型的方案在较少语义场景下和光度变原创 2020-09-11 18:55:40 · 288 阅读 · 1 评论 -
A Point-Line Feature based Visual SLAM Method in Dynamic Indoor Scene
A Point-Line Feature based Visual SLAM Method in Dynamic Indoor SceneAbstractIntroductionRelated WorkFeature Extraction and MatchingDynamic Objects EliminationMotion EstimationExperimentConclusion and Future Work本文主要的创新点在于将特征点和线性特征信息相结合,同时通过对动态目标上特征点的踢除避免原创 2020-09-08 22:35:47 · 325 阅读 · 0 评论 -
PL-SVO: Semi-Direct Monocular Visual Odometry by Combining Points and Line Segments简介
PL-SVO: Semi-Direct Monocular Visual Odometry by Combining Points and Line SegmentsAbstractIntroductionSystem OverviewSemi-Direct Monocular Visual OdometryMappingExperimental ValidationConclusionsAbstract本文是基于之前的半稠密视觉slam系统SVO进行的改进,在此基础上添加线性特征信息使得整个系统在相对原创 2020-09-08 15:33:36 · 535 阅读 · 0 评论 -
EIP-VIO: Edge-Induced Points based Monocular Visual-Inertial Odometry
EIP-VIO: Edge-Induced Points based Monocular Visual-Inertial OdometryAbstractIntroductionThe Proposed MethodExperimental ResultConclusionsAbstract本文提出一种通过线性特征信息引导高区分度点特征的直接法结合IMU测量单元的视觉里程计系统。该系统相比传统的直接法而言对像素点位置的跟踪更加精确、稳定,整个系统的鲁棒性更高。Introduction该部分将传统的视原创 2020-09-06 20:59:44 · 309 阅读 · 0 评论 -
Direct Monocular Odometry Using Points and Lines
Direct Monocular Odometry Using Points and LinesAbstractIntroductionRelated WorkProblem DescriptionTrackingMappingExperiments and ResultConclusionsAckonwledgmentsAbstract本文提出了一种结合特征信息和直接法的视觉里程计方式,其中特征信息不只有单纯的点特征还有edge类型的特征信息,使得系统更加鲁棒且对于光照变换适应性更好。通过对像素亮度误原创 2020-09-05 22:51:23 · 307 阅读 · 0 评论 -
Accurate stereo Visual Odometry based on keypoint selection
Accurate stereo Visual Odometry based on keypoint selectionAbstractIntroductionRelated WorksMethodologyEvaluationConclusionAcknowledgmentAbstract本文提出了一种基于循环匹配特征点的双目视觉里程计方法,同时在每一次双目帧之间通过INSADINSADINSAD和极线几何知识进行特征点之间的匹配,在帧与帧之间通过KLTKLTKLT的方法实现特征点的跟踪,这样使得以特征原创 2020-09-02 22:19:31 · 183 阅读 · 0 评论 -
CV-SLAM using Line and Point Features
CV-SLAM using Line and Point FeaturesAbstractIntroductionEKF-SLAM FrameworkFeature ExtractionExperimental ResultsConclusionAcknowledgmentAbstract本文主要针对视觉里程计在室内的运用,选择的场景是天花板,结合多种类型的特征信息进行移动机器人的定位和轨迹复现。Introduction在基于特征类型的视觉里程计中有效特征的提取以及特征数据的一致性都是研究的重点内容原创 2020-09-02 16:04:16 · 262 阅读 · 0 评论 -
MEVO: Multi-Environment Stereo Visual Odometry
MEVO: Multi-Environment Stereo Visual OdometryAbstractIntroductionFeature MatchingMotion EstimationExperimentsConclusionsAbstract在某些场景中语义信息不是很丰富,单纯的依靠点型特征信息无法实现鲁棒性的获取位姿变换信息,因此本文提出一种结合点特征和线性特征的视觉里程计,提升了视觉里程计的鲁棒性,同时在线性特征匹配中对右侧视图中的特征部分由于遮挡等原因无法完全可视化,依旧可以在右侧原创 2020-08-27 15:01:09 · 208 阅读 · 1 评论