Bio-inspired visual systems have garnered significant attention in robotics owing to their energy efficiency,rapid dynamic response,and environmental adaptability.Among these,event cameras-bio-inspired sensors that as...Bio-inspired visual systems have garnered significant attention in robotics owing to their energy efficiency,rapid dynamic response,and environmental adaptability.Among these,event cameras-bio-inspired sensors that asynchronously report pixel-level brightness changes called’events’,stand out because of their ability to capture dynamic changes with minimal energy consumption,making them suitable for challenging conditions,such as low light or high-speed motion.However,current mapping and localization methods for event cameras depend primarily on point and line features,which struggle in sparse or low-feature environments and are unsuitable for static or slow-motion scenarios.We addressed these challenges by proposing a bio-inspired vision mapping and localization method using active LED markers(ALMs)combined with reprojection error optimization and asynchronous Kalman fusion.Our approach replaces traditional features with ALMs,thereby enabling accurate tracking under dynamic and low-feature conditions.The global mapping accuracy significantly improved by minimizing the reprojection error,with corner errors reduced from 16.8 cm to 3.1 cm after 400 iterations.The asynchronous Kalman fusion of multiple camera pose estimations from ALMs ensures precise localization with a high temporal efficiency.This method achieved a mean translation error of 0.078 m and a rotational error of 5.411°while evaluating dynamic motion.In addition,the method supported an output rate of 4.5 kHz while maintaining high localization accuracy in UAV spiral flight experiments.These results demonstrate the potential of the proposed approach for real-time robot localization in challenging environments.展开更多
To enhance the image motion compensation accuracy of off-axis three-mirror anastigmatic( TMA)three-line array aerospace mapping cameras,a new method of image motion velocity field modeling is proposed in this paper. F...To enhance the image motion compensation accuracy of off-axis three-mirror anastigmatic( TMA)three-line array aerospace mapping cameras,a new method of image motion velocity field modeling is proposed in this paper. Firstly,based on the imaging principle of mapping cameras,an analytical expression of image motion velocity of off-axis TMA three-line array aerospace mapping cameras is deduced from different coordinate systems we established and the attitude dynamics principle. Then,the case of a three-line array mapping camera is studied,in which the simulation of the focal plane image motion velocity fields of the forward-view camera,the nadir-view camera and the backward-view camera are carried out,and the optimization schemes for image motion velocity matching and drift angle matching are formulated according the simulation results. Finally,this method is verified with a dynamic imaging experimental system. The results are indicative of that when image motion compensation for nadir-view camera is conducted using the proposed image motion velocity field model,the line pair of target images at Nyquist frequency is clear and distinguishable. Under the constraint that modulation transfer function( MTF) reduces by 5%,when the horizontal frequencies of the forward-view camera and the backward-view camera are adjusted uniformly according to the proposed image motion velocity matching scheme,the time delay integration( TDI) stages reach 6 at most. When the TDI stages are more than 6,the three groups of camera will independently undergo horizontal frequency adjustment. However, when the proposed drift angle matching scheme is adopted for uniform drift angle adjustment,the number of TDI stages will not exceed 81. The experimental results have demonstrated the validity and accuracy of the proposed image motion velocity field model and matching optimization scheme,providing reliable basis for on-orbit image motion compensation of aerospace mapping cameras.展开更多
The geometric accuracy of topographic mapping with high-resolution remote sensing images is inevita-bly affected by the orbiter attitude jitter.Therefore,it is necessary to conduct preliminary research on the stereo m...The geometric accuracy of topographic mapping with high-resolution remote sensing images is inevita-bly affected by the orbiter attitude jitter.Therefore,it is necessary to conduct preliminary research on the stereo mapping camera equipped on lunar orbiter before launching.In this work,an imaging simulation method consid-ering the attitude jitter is presented.The impact analysis of different attitude jitter on terrain undulation is conduct-ed by simulating jitter at three attitude angles,respectively.The proposed simulation method is based on the rigor-ous sensor model,using the lunar digital elevation model(DEM)and orthoimage as reference data.The orbit and attitude of the lunar stereo mapping camera are simulated while considering the attitude jitter.Two-dimensional simulated stereo images are generated according to the position and attitude of the orbiter in a given orbit.Experi-mental analyses were conducted by the DEM with the simulated stereo image.The simulation imaging results demonstrate that the proposed method can ensure imaging efficiency without losing the accuracy of topographic mapping.The effect of attitude jitter on the stereo mapping accuracy of the simulated images was analyzed through a DEM comparison.展开更多
An effective approach,mapping the texture for building model based on the digital photogrammetric theory,is proposed.The easily-acquired image sequences from digital video camera on helicopter are used as texture reso...An effective approach,mapping the texture for building model based on the digital photogrammetric theory,is proposed.The easily-acquired image sequences from digital video camera on helicopter are used as texture resource,and the correspondence between the space edge in building geometry model and its line feature in image sequences is determined semi-automatically.The experimental results in production of three-dimensional data for car navigation show us an attractive future both in efficiency and effect.展开更多
Although deep learning methods have been widely applied in slam visual odometry(VO)over the past decade with impressive improvements,the accuracy remains limited in complex dynamic environments.In this paper,a composi...Although deep learning methods have been widely applied in slam visual odometry(VO)over the past decade with impressive improvements,the accuracy remains limited in complex dynamic environments.In this paper,a composite mask-based generative adversarial network(CMGAN)is introduced to predict camera motion and binocular depth maps.Specifically,a perceptual generator is constructed to obtain the corresponding parallax map and optical flow between two neighboring frames.Then,an iterative pose improvement strategy is proposed to improve the accuracy of pose estimation.Finally,a composite mask is embedded in the discriminator to sense structural deformation in the synthesized virtual image,thereby increasing the overall structural constraints of the network model,improving the accuracy of camera pose estimation,and reducing drift issues in the VO.Detailed quantitative and qualitative evaluations on the KITTI dataset show that the proposed framework outperforms existing conventional,supervised learning and unsupervised depth VO methods,providing better results in both pose estimation and depth estimation.展开更多
The technique of imaging or tracking objects outside the field of view(FOV)through a reflective relay surface,usually called non-line-of-sight(NLOS)imaging,has been a popular research topic in recent years.Although NL...The technique of imaging or tracking objects outside the field of view(FOV)through a reflective relay surface,usually called non-line-of-sight(NLOS)imaging,has been a popular research topic in recent years.Although NLOS imaging can be achieved through methods such as detector design,optical path inverse operation algorithm design,or deep learning,challenges such as high costs,complex algorithms,and poor results remain.This study introduces a simple algorithm-based rapid depth imaging device,namely,the continuous-wave time-offlight range imaging camera(CW-TOF camera),to address the decoupled imaging challenge of differential scattering characteristics in an object-relay surface by quantifying the differential scattering signatures through statistical analysis of light propagation paths.A scalable scattering mapping(SSM)theory has been proposed to explain the degradation process of clear images.High-quality NLOS object 3D imaging has been achieved through a data-driven approach.To verify the effectiveness of the proposed algorithm,experiments were conducted using an optical platform and real-world scenarios.The objects on the optical platform include plaster sculptures and plastic letters,while relay surfaces consist of polypropylene(PP)plastic boards,acrylic boards,and standard Lambertian diffusers.In real-world scenarios,the object is clothing,with relay surfaces including painted doors and white plaster walls.Imaging data were collected for different combinations of objects and relay surfaces for training and testing,totaling 210,000 depth images.The reconstruction of NLOS images in the laboratory and real-world is excellent according to subjective evaluation;thus,our approach can realize NLOS imaging in harsh natural scenes and advances the practical application of NLOS imaging.展开更多
This paper addresses a sensor-based simultaneous localization and mapping （SLAM） algorithm for camera tracking in a virtual studio environment. The traditional camera tracking methods in virtual studios are vision-b...This paper addresses a sensor-based simultaneous localization and mapping （SLAM） algorithm for camera tracking in a virtual studio environment. The traditional camera tracking methods in virtual studios are vision-based or sensor-based. However, the chroma keying process in virtual studios requires color cues, such as blue background, to segment foreground objects to be inserted into images and videos. Chroma keying limits the application of vision-based tracking methods in virtual studios since the background cannot provide enough feature information. Furthermore, the conventional sensor-based tracking approaches suffer from the jitter, drift or expensive computation due to the characteristics of individual sensor system. Therefore, the SLAM techniques from the mobile robot area are first investigated and adapted to the camera tracking area. Then, a sensor-based SLAM extension algorithm for two dimensional （2D） camera tracking in virtual studio is described. Also, a technique called map adjustment is proposed to increase the accuracy＇ and efficiency of the algorithm. The feasibility and robustness of the algorithm is shown by experiments. The simulation results demonstrate that the sensor-based SLAM algorithm can satisfy the fundamental 2D camera tracking requirement in virtual studio environment.展开更多
巡检机器人对室内场景进行自主导航监测时,采用视觉同时定位与地图构建(simultaneous localization and mapping,SLAM)方法构建的三维深度地图存在实时性不高、定位精度下降的问题。对此,提出了一种基于RGB-D相机和优化RTAB-Map(real ti...巡检机器人对室内场景进行自主导航监测时,采用视觉同时定位与地图构建(simultaneous localization and mapping,SLAM)方法构建的三维深度地图存在实时性不高、定位精度下降的问题。对此,提出了一种基于RGB-D相机和优化RTAB-Map(real time appearance based mapping)算法的巡检机器人视觉导航方法。首先,通过重新配置RTAB-Map点云更新频率,实现算法优化,构建稠密的点云地图后;采用启发式A*算法、动态窗口法(dynamic window approach,DWA)分别制定全局与局部巡检路径,通过自适应蒙特卡罗定位(adaptive Monte Carlo localization,AMCL)方法更新机器人的实时位姿信息,再将搭建好的实体巡检机器人在软件、硬件平台上完成视觉导航测试实验。结果表明:优化后的RTAB-Map算法运行时的内存占比稍有增加,但获得与真实环境一致性更高的三维深度地图,在一定程度上提高视觉导航的准确性与实用性。展开更多
Various spacecraft and satellites from the world’s best space agencies are exploring Mars since 1970, constantly with great ability to capture the maximum amount of dataset for a better understanding of the red plane...Various spacecraft and satellites from the world’s best space agencies are exploring Mars since 1970, constantly with great ability to capture the maximum amount of dataset for a better understanding of the red planet. In this paper, we propose a new method for making a mosaic of Mars Reconnaissance Orbiter (MRO) spacecraft payload Context Camera (CTX) images. In this procedure, we used ERDAS Imagine for image rectification and mosaicking as a tool for image processing, which is a new and unique method of generating a mosaic of thousands of CTX images to visualize the large-scale areas. The output product will be applicable for mapping of Martian geomorphological features, 2D mapping of the linear feature with high resolution, crater counting, and morphometric analysis to a certain extent.展开更多
Real-time indoor camera localization is a significant problem in indoor robot navigation and surveillance systems.The scene can change during the image sequence and plays a vital role in the localization performance o...Real-time indoor camera localization is a significant problem in indoor robot navigation and surveillance systems.The scene can change during the image sequence and plays a vital role in the localization performance of robotic applications in terms of accuracy and speed.This research proposed a real-time indoor camera localization system based on a recurrent neural network that detects scene change during the image sequence.An annotated image dataset trains the proposed system and predicts the camera pose in real-time.The system mainly improved the localization performance of indoor cameras by more accurately predicting the camera pose.It also recognizes the scene changes during the sequence and evaluates the effects of these changes.This system achieved high accuracy and real-time performance.The scene change detection process was performed using visual rhythm and the proposed recurrent deep architecture,which performed camera pose prediction and scene change impact evaluation.Overall,this study proposed a novel real-time localization system for indoor cameras that detects scene changes and shows how they affect localization performance.展开更多
在动态场景下,视觉同时定位与地图构建(simultaneous localization and mapping, SLAM)通常与深度学习方法结合提高系统的定位精度.针对深度学习方法运行时产生的时间延迟,导致系统难以达到流式处理要求的问题,提出一种面向动态场景下视...在动态场景下,视觉同时定位与地图构建(simultaneous localization and mapping, SLAM)通常与深度学习方法结合提高系统的定位精度.针对深度学习方法运行时产生的时间延迟,导致系统难以达到流式处理要求的问题,提出一种面向动态场景下视觉SLAM的流感知定位方法.首先针对传统评估指标只考虑定位精度的问题,提出流式评估指标,该指标同时考虑定位精度和时间延迟,能够准确反映系统的流式处理性能;其次针对传统视觉SLAM方法无法实现流式处理的问题,提出流感知的视觉定位方法,通过多线程并行和相机位姿预测相结合的方式,获得持续稳定的相机位姿输出.在BONN数据集和真实场景上的实验结果表明,所提方法能够有效地提升动态场景下采用深度学习方法的视觉定位的流性能.基于BONN数据集和流式评估方式的评估结果表明,与DynaSLAM方法对比,所提方法的绝对轨迹误差(APE),相对平移误差(RPE_trans)和相对旋转误差(RPE_angle)分别下降80.438%, 56.180%和54.676%.在真实场景下的实验结果表明,所提方法可以得到与实际相符的相机轨迹.展开更多
同步定位与建图(simultaneous localization and mapping,SLAM)技术能够帮助移动机器人在没有先验信息的条件下,为其提供地图和自身位置信息,已成为移动机器人自主导航的主流解决方案,其中以相机为传感器的视觉SLAM,有着体积小巧、成本...同步定位与建图(simultaneous localization and mapping,SLAM)技术能够帮助移动机器人在没有先验信息的条件下,为其提供地图和自身位置信息,已成为移动机器人自主导航的主流解决方案,其中以相机为传感器的视觉SLAM,有着体积小巧、成本低、高分辨率等优势。随着研究者们对SLAM问题的深入研究,SLAM领域相关成果已非常丰富,但是有关视觉场景下的SLAM论述还不够系统。文中首先介绍了视觉SLAM的基本原理,之后对于传统视觉SLAM与基于深度学习的视觉SLAM两个方面阐述了视觉SLAM的研究方法,从地图类型以及特点等方面进行对比分析,为移动机器人的视觉SLAM技术研究提供了参考。展开更多
基金Supported by Beijing Natural Science Foundation(Grant No.L231004)Young Elite Scientists Sponsorship Program by CAST(Grant No.2022QNRC001)+2 种基金Fundamental Research Funds for the Central Universities(Grant No.2025JBMC039)National Key Research and Development Program(Grant No.2022YFC2805200)National Natural Science Foundation of China(Grant No.52371338).
文摘Bio-inspired visual systems have garnered significant attention in robotics owing to their energy efficiency,rapid dynamic response,and environmental adaptability.Among these,event cameras-bio-inspired sensors that asynchronously report pixel-level brightness changes called’events’,stand out because of their ability to capture dynamic changes with minimal energy consumption,making them suitable for challenging conditions,such as low light or high-speed motion.However,current mapping and localization methods for event cameras depend primarily on point and line features,which struggle in sparse or low-feature environments and are unsuitable for static or slow-motion scenarios.We addressed these challenges by proposing a bio-inspired vision mapping and localization method using active LED markers(ALMs)combined with reprojection error optimization and asynchronous Kalman fusion.Our approach replaces traditional features with ALMs,thereby enabling accurate tracking under dynamic and low-feature conditions.The global mapping accuracy significantly improved by minimizing the reprojection error,with corner errors reduced from 16.8 cm to 3.1 cm after 400 iterations.The asynchronous Kalman fusion of multiple camera pose estimations from ALMs ensures precise localization with a high temporal efficiency.This method achieved a mean translation error of 0.078 m and a rotational error of 5.411°while evaluating dynamic motion.In addition,the method supported an output rate of 4.5 kHz while maintaining high localization accuracy in UAV spiral flight experiments.These results demonstrate the potential of the proposed approach for real-time robot localization in challenging environments.
基金Sponsored by the National High Technology Research and Development Program of China(Grant No.863-2-5-1-13B)the Jilin Province Science and Technology Development Plan Item(Grant No.20130522107JH)
文摘To enhance the image motion compensation accuracy of off-axis three-mirror anastigmatic( TMA)three-line array aerospace mapping cameras,a new method of image motion velocity field modeling is proposed in this paper. Firstly,based on the imaging principle of mapping cameras,an analytical expression of image motion velocity of off-axis TMA three-line array aerospace mapping cameras is deduced from different coordinate systems we established and the attitude dynamics principle. Then,the case of a three-line array mapping camera is studied,in which the simulation of the focal plane image motion velocity fields of the forward-view camera,the nadir-view camera and the backward-view camera are carried out,and the optimization schemes for image motion velocity matching and drift angle matching are formulated according the simulation results. Finally,this method is verified with a dynamic imaging experimental system. The results are indicative of that when image motion compensation for nadir-view camera is conducted using the proposed image motion velocity field model,the line pair of target images at Nyquist frequency is clear and distinguishable. Under the constraint that modulation transfer function( MTF) reduces by 5%,when the horizontal frequencies of the forward-view camera and the backward-view camera are adjusted uniformly according to the proposed image motion velocity matching scheme,the time delay integration( TDI) stages reach 6 at most. When the TDI stages are more than 6,the three groups of camera will independently undergo horizontal frequency adjustment. However, when the proposed drift angle matching scheme is adopted for uniform drift angle adjustment,the number of TDI stages will not exceed 81. The experimental results have demonstrated the validity and accuracy of the proposed image motion velocity field model and matching optimization scheme,providing reliable basis for on-orbit image motion compensation of aerospace mapping cameras.
基金Supported by the National Natural Science Foundation of China(42221002,42171432)Shanghai Municipal Science and Technology Major Project(2021SHZDZX0100)the Fundamental Research Funds for the Central Universities.
文摘The geometric accuracy of topographic mapping with high-resolution remote sensing images is inevita-bly affected by the orbiter attitude jitter.Therefore,it is necessary to conduct preliminary research on the stereo mapping camera equipped on lunar orbiter before launching.In this work,an imaging simulation method consid-ering the attitude jitter is presented.The impact analysis of different attitude jitter on terrain undulation is conduct-ed by simulating jitter at three attitude angles,respectively.The proposed simulation method is based on the rigor-ous sensor model,using the lunar digital elevation model(DEM)and orthoimage as reference data.The orbit and attitude of the lunar stereo mapping camera are simulated while considering the attitude jitter.Two-dimensional simulated stereo images are generated according to the position and attitude of the orbiter in a given orbit.Experi-mental analyses were conducted by the DEM with the simulated stereo image.The simulation imaging results demonstrate that the proposed method can ensure imaging efficiency without losing the accuracy of topographic mapping.The effect of attitude jitter on the stereo mapping accuracy of the simulated images was analyzed through a DEM comparison.
文摘An effective approach,mapping the texture for building model based on the digital photogrammetric theory,is proposed.The easily-acquired image sequences from digital video camera on helicopter are used as texture resource,and the correspondence between the space edge in building geometry model and its line feature in image sequences is determined semi-automatically.The experimental results in production of three-dimensional data for car navigation show us an attractive future both in efficiency and effect.
基金supported by the Program of Graduate Education and Teaching Reform in Tianjin University of Technology(Nos.YBXM2204 and ZDXM2202)the National Natural Science Foundation of China(Nos.62203331 and 62103299)。
文摘Although deep learning methods have been widely applied in slam visual odometry(VO)over the past decade with impressive improvements,the accuracy remains limited in complex dynamic environments.In this paper,a composite mask-based generative adversarial network(CMGAN)is introduced to predict camera motion and binocular depth maps.Specifically,a perceptual generator is constructed to obtain the corresponding parallax map and optical flow between two neighboring frames.Then,an iterative pose improvement strategy is proposed to improve the accuracy of pose estimation.Finally,a composite mask is embedded in the discriminator to sense structural deformation in the synthesized virtual image,thereby increasing the overall structural constraints of the network model,improving the accuracy of camera pose estimation,and reducing drift issues in the VO.Detailed quantitative and qualitative evaluations on the KITTI dataset show that the proposed framework outperforms existing conventional,supervised learning and unsupervised depth VO methods,providing better results in both pose estimation and depth estimation.
基金National Key Research and Development Program of China(2023YFC3321600)Special Project for Research and Development in Key Areas of Guangdong Province(2023ZDZX1044)+1 种基金Zhuhai Multimodal Intelligent Vision Engineering Technology Research Center(2320004002292)Zhuhai Basic and Applied Basic Research Foundation(2220004002937)。
文摘The technique of imaging or tracking objects outside the field of view(FOV)through a reflective relay surface,usually called non-line-of-sight(NLOS)imaging,has been a popular research topic in recent years.Although NLOS imaging can be achieved through methods such as detector design,optical path inverse operation algorithm design,or deep learning,challenges such as high costs,complex algorithms,and poor results remain.This study introduces a simple algorithm-based rapid depth imaging device,namely,the continuous-wave time-offlight range imaging camera(CW-TOF camera),to address the decoupled imaging challenge of differential scattering characteristics in an object-relay surface by quantifying the differential scattering signatures through statistical analysis of light propagation paths.A scalable scattering mapping(SSM)theory has been proposed to explain the degradation process of clear images.High-quality NLOS object 3D imaging has been achieved through a data-driven approach.To verify the effectiveness of the proposed algorithm,experiments were conducted using an optical platform and real-world scenarios.The objects on the optical platform include plaster sculptures and plastic letters,while relay surfaces consist of polypropylene(PP)plastic boards,acrylic boards,and standard Lambertian diffusers.In real-world scenarios,the object is clothing,with relay surfaces including painted doors and white plaster walls.Imaging data were collected for different combinations of objects and relay surfaces for training and testing,totaling 210,000 depth images.The reconstruction of NLOS images in the laboratory and real-world is excellent according to subjective evaluation;thus,our approach can realize NLOS imaging in harsh natural scenes and advances the practical application of NLOS imaging.
文摘This paper addresses a sensor-based simultaneous localization and mapping （SLAM） algorithm for camera tracking in a virtual studio environment. The traditional camera tracking methods in virtual studios are vision-based or sensor-based. However, the chroma keying process in virtual studios requires color cues, such as blue background, to segment foreground objects to be inserted into images and videos. Chroma keying limits the application of vision-based tracking methods in virtual studios since the background cannot provide enough feature information. Furthermore, the conventional sensor-based tracking approaches suffer from the jitter, drift or expensive computation due to the characteristics of individual sensor system. Therefore, the SLAM techniques from the mobile robot area are first investigated and adapted to the camera tracking area. Then, a sensor-based SLAM extension algorithm for two dimensional （2D） camera tracking in virtual studio is described. Also, a technique called map adjustment is proposed to increase the accuracy＇ and efficiency of the algorithm. The feasibility and robustness of the algorithm is shown by experiments. The simulation results demonstrate that the sensor-based SLAM algorithm can satisfy the fundamental 2D camera tracking requirement in virtual studio environment.
文摘巡检机器人对室内场景进行自主导航监测时,采用视觉同时定位与地图构建(simultaneous localization and mapping,SLAM)方法构建的三维深度地图存在实时性不高、定位精度下降的问题。对此,提出了一种基于RGB-D相机和优化RTAB-Map(real time appearance based mapping)算法的巡检机器人视觉导航方法。首先,通过重新配置RTAB-Map点云更新频率,实现算法优化,构建稠密的点云地图后;采用启发式A*算法、动态窗口法(dynamic window approach,DWA)分别制定全局与局部巡检路径,通过自适应蒙特卡罗定位(adaptive Monte Carlo localization,AMCL)方法更新机器人的实时位姿信息,再将搭建好的实体巡检机器人在软件、硬件平台上完成视觉导航测试实验。结果表明:优化后的RTAB-Map算法运行时的内存占比稍有增加,但获得与真实环境一致性更高的三维深度地图,在一定程度上提高视觉导航的准确性与实用性。
文摘Various spacecraft and satellites from the world’s best space agencies are exploring Mars since 1970, constantly with great ability to capture the maximum amount of dataset for a better understanding of the red planet. In this paper, we propose a new method for making a mosaic of Mars Reconnaissance Orbiter (MRO) spacecraft payload Context Camera (CTX) images. In this procedure, we used ERDAS Imagine for image rectification and mosaicking as a tool for image processing, which is a new and unique method of generating a mosaic of thousands of CTX images to visualize the large-scale areas. The output product will be applicable for mapping of Martian geomorphological features, 2D mapping of the linear feature with high resolution, crater counting, and morphometric analysis to a certain extent.
文摘Real-time indoor camera localization is a significant problem in indoor robot navigation and surveillance systems.The scene can change during the image sequence and plays a vital role in the localization performance of robotic applications in terms of accuracy and speed.This research proposed a real-time indoor camera localization system based on a recurrent neural network that detects scene change during the image sequence.An annotated image dataset trains the proposed system and predicts the camera pose in real-time.The system mainly improved the localization performance of indoor cameras by more accurately predicting the camera pose.It also recognizes the scene changes during the sequence and evaluates the effects of these changes.This system achieved high accuracy and real-time performance.The scene change detection process was performed using visual rhythm and the proposed recurrent deep architecture,which performed camera pose prediction and scene change impact evaluation.Overall,this study proposed a novel real-time localization system for indoor cameras that detects scene changes and shows how they affect localization performance.
文摘同步定位与建图(simultaneous localization and mapping,SLAM)技术能够帮助移动机器人在没有先验信息的条件下,为其提供地图和自身位置信息,已成为移动机器人自主导航的主流解决方案,其中以相机为传感器的视觉SLAM,有着体积小巧、成本低、高分辨率等优势。随着研究者们对SLAM问题的深入研究,SLAM领域相关成果已非常丰富,但是有关视觉场景下的SLAM论述还不够系统。文中首先介绍了视觉SLAM的基本原理,之后对于传统视觉SLAM与基于深度学习的视觉SLAM两个方面阐述了视觉SLAM的研究方法,从地图类型以及特点等方面进行对比分析,为移动机器人的视觉SLAM技术研究提供了参考。
