Results 1 - 10
of
14
Live Tracking and Mapping from Both General and Rotation-Only Camera Motion
"... We present an approach to real-time tracking and mapping that supports any type of camera motion in 3D environments, that is, general (parallax-inducing) as well as rotation-only (degenerate) motions. Our approach effectively generalizes both a panorama mapping and tracking system and a keyframe-bas ..."
Abstract
-
Cited by 6 (1 self)
- Add to MetaCart
(Show Context)
We present an approach to real-time tracking and mapping that supports any type of camera motion in 3D environments, that is, general (parallax-inducing) as well as rotation-only (degenerate) motions. Our approach effectively generalizes both a panorama mapping and tracking system and a keyframe-based Simultaneous Localization and Mapping (SLAM) system, behaving like one or the other depending on the camera movement. It seamlessly switches between the two and is thus able to track and map through arbitrary sequences of general and rotation-only camera movements. Key elements of our approach are to design each system component such that it is compatible with both panoramic data and Structure-from-Motion data, and the use of the ‘Geometric Robust Information Criterion ’ to decide whether the transformation between a given pair of frames can best be modeled with an essential matrix E, or with a homography H. Further key features are that no separate initialization step is needed, that the reconstruction is unbiased, and that the system continues to collect and map data after tracking failure, thus creating separate tracks which are later merged if they overlap. The latter is in contrast to most existing tracking and mapping systems, which suspend tracking and mapping, thus discarding valuable data, while trying to relocalize the camera with respect to the initial map. We tested our system on a variety of video sequences, successfully tracking through different camera motions and fully automatically building panoramas as well as 3D structures. 1
Temporal Calibration in Multisensor Tracking Setups
"... Spatial tracking is one of the most challenging parts of Augmented Reality. Many AR applications rely on the fusion of several tracking systems in order to optimize the overall performance. While the topic of sensor fusion has already seen considerable interest, most results only deal with the integ ..."
Abstract
-
Cited by 6 (3 self)
- Add to MetaCart
Spatial tracking is one of the most challenging parts of Augmented Reality. Many AR applications rely on the fusion of several tracking systems in order to optimize the overall performance. While the topic of sensor fusion has already seen considerable interest, most results only deal with the integration of particular setups. A crucial part of sensor fusion is the temporal alignment of the sensor signals, as sensors in general are not synchronized. We present a general method to calibrate the temporal offset between different sensors by applying the normalized cross correlation method.
Benchmarking Inertial Sensor-Aided Localization and Tracking Methods
"... benchmark datasets, that do not contain any inertial sensor measurements. At the example of metaio’s template tracking benchmarking set [18], we show how synthesizing inertial sensor measurements from the ground truth poses enables the evaluation of such methods at virtually no extra cost while prov ..."
Abstract
-
Cited by 4 (1 self)
- Add to MetaCart
(Show Context)
benchmark datasets, that do not contain any inertial sensor measurements. At the example of metaio’s template tracking benchmarking set [18], we show how synthesizing inertial sensor measurements from the ground truth poses enables the evaluation of such methods at virtually no extra cost while providing comparable results to using real inertial sensor data, which was validated using the setup in the right photo. This paper investigates means to benchmark methods for camera pose localization and tracking that in addition to a camera image make use of inertial sensor measurements. In particular the direction of the gravity has recently shown to provide useful information to aid vision-based approaches making them outperform visiononly methods. Obviously, it is desirable to benchmark the performance of such methods and to compare them with state-of-the-art approaches, but to the best of our knowledge, all publicly available benchmarking datasets unfortunately lack gravity information. We present different simple means to generate one’s own benchmarks for inertial sensor-aided localization and tracking methods and most considerably show how existing datasets, that do not have inertial sensor data, can be exploited. We demonstrate how to evaluate Gravity-Aligned Feature Descriptors (GAFD) and Gravity-Rectified Feature Descriptors (GREFD) on an existing benchmark dataset with ground truth poses. By synthesizing gravity measurements from these poses we achieve similar results to using real sensor measurements at significantly less effort. Most importantly, the proposed procedure enables the comparison with existing evaluation results on the same data. The paper concludes with a requirements analysis and suggestions for the design of future benchmarking datasets for localization and tracking methods. 1
Outdoor Mobile Localization from Panoramic Imagery
"... We describe an end-to-end system for mobile, vision-based localization and tracking in urban environments. Our system uses panoramic imagery which is processed and indexed to provide localization coverage over a large area using few capture points. We utilize a client-server model which allows for r ..."
Abstract
-
Cited by 1 (1 self)
- Add to MetaCart
(Show Context)
We describe an end-to-end system for mobile, vision-based localization and tracking in urban environments. Our system uses panoramic imagery which is processed and indexed to provide localization coverage over a large area using few capture points. We utilize a client-server model which allows for remote computation and data storage while maintaining real-time tracking performance. Previous search results are cached and re-used by the mobile client to minimize communication overhead. We evaluate the use of the system for flexible real-time camera tracking in large outdoor spaces.
Evaluating the Impact of Recovery Density on Augmented Reality Tracking
"... Natural feature tracking systems for augmented reality are highly accurate, but can suffer from lost tracking. When registration is lost, the system must be able to re-localize and recover tracking. Likewise, when a camera is new to a scene, it must be able to perform the related task of localizatio ..."
Abstract
-
Cited by 1 (1 self)
- Add to MetaCart
(Show Context)
Natural feature tracking systems for augmented reality are highly accurate, but can suffer from lost tracking. When registration is lost, the system must be able to re-localize and recover tracking. Likewise, when a camera is new to a scene, it must be able to perform the related task of localization. Localization and re-localization can only be performed at certain points or when viewing particular objects or parts of the scene with a sufficient number and quality of recognizable features to allow for tracking recovery. We explore how the density of such recovery locations/poses influences the time it takes users to resume tracking. We focus our evaluation on two generalized techniques for localization: keyframe-based and model-based. For the keyframe-based approach we assume a constant collection rate for keyframes. We find that at practical collection rates, the task of localization to a previously acquired keyframe that is shown to the user does not become more time-consuming as the interval between keyframes increases. For a localization approach using model data, we consider a grid of points around the model at which localization is guaranteed to succeed. We find that the user interface is crucial to successful localization. Localization can occur quickly if users do not need to orient themselves to marked localization points. When users are forced to mentally register themselves with a map of the scene, localization quickly becomes impractical as the distance to the next localization point increases. We contend that our results will help future designers of localization techniques to better plan for the effects of their proposed solutions.
Pixel-Wise Closed-Loop Registration in Video-Based Augmented Reality
"... Figure 1: Comparison between a conventional open-loop registration approach (a) and our closed-loop registration approach employing both world-space pose refinement (b) and screen-space pixel-wise corrections (c). The dragon, the square & axes object and shadows are augmented. In (a), errors in ..."
Abstract
-
Cited by 1 (1 self)
- Add to MetaCart
(Show Context)
Figure 1: Comparison between a conventional open-loop registration approach (a) and our closed-loop registration approach employing both world-space pose refinement (b) and screen-space pixel-wise corrections (c). The dragon, the square & axes object and shadows are augmented. In (a), errors in estimates of camera intrinsics and extrinsics (6DOF pose) result in visible misregistration that is neither measured nor corrected as part of a conventional open-loop approach. Such registration errors include direct virtual object misregistration (e.g., the square & axes object), and “phantom ” object misregistration errors including incorrect real-to-virtual occlusions (e.g., between the tower and the dragon) and associated shading effects between the real and the virtual (e.g., virtual shadow cast by the tower). See the “zoomed in ” portions of the images. In our closed-loop approach, registration errors are detected using a model of the real scene, and corrected in both world space using camera pose refinement (b) and screen space using pixel-wise corrections (c) to address both rigid and non-rigid registration errors. The final result (c) is spatially accurate and exhibits visually coherent registration. In Augmented Reality (AR), visible misregistration can be caused by many inherent error sources, such as errors in tracking, calibration, and modeling. In this paper we present a novel pixel-wise closed-loop registration framework that can automatically detect and correct registration errors using a reference model comprised
DT-SLAM: Deferred Triangulation for Robust SLAM
"... Obtaining a good baseline between different video frames is one of the key elements in vision-based monocular SLAM systems. However, if the video frames contain only a few 2D feature correspondences with a good base-line, or the camera only rotates without sufficient translation in the beginning, tr ..."
Abstract
- Add to MetaCart
Obtaining a good baseline between different video frames is one of the key elements in vision-based monocular SLAM systems. However, if the video frames contain only a few 2D feature correspondences with a good base-line, or the camera only rotates without sufficient translation in the beginning, tracking and mapping becomes un-stable. We introduce a real-time visual SLAM system that incrementally tracks individual 2D features, and estimates camera pose by using matched 2D features, regardless of the length of the baseline. Triangulating 2D features into 3D points is deferred until keyframes with sufficient base-line for the features are available. Our method can also deal with pure rotational motions, and fuse the two types of measurements in a bundle adjustment step. Adaptive criteria for keyframe selection are also introduced for efficient optimization and dealing with multiple maps. We demonstrate that our SLAM system improves camera pose estimates and robustness, even with purely rotational motions.
Incremental Dense Reconstruction from Sparse 3D Points with an Integrated Level-of-Detail Concept
"... Abstract. For decades scene reconstruction from multiple images is a topic in computer vision and photogrammetry communities. Typical applications require very precise reconstructions and are not bound to a limited computation time. Techniques for these applications are based on complete sets of ima ..."
Abstract
- Add to MetaCart
(Show Context)
Abstract. For decades scene reconstruction from multiple images is a topic in computer vision and photogrammetry communities. Typical applications require very precise reconstructions and are not bound to a limited computation time. Techniques for these applications are based on complete sets of images to compute the scene geometry. They require a huge amount of resources and computation time before delivering results for visualization or further processing. In the application of disaster management these approaches are not an option since the reconstructed data has to be available as soon as possible. Especially, when it comes to Miniature Unmanned Aerial Vehicles (MUAVs) sending aerial images to a ground station wirelessly while flying, operators can use the 3D data to explore the virtual world and to control the MUAVs. In this paper an incremental approach for dense reconstructions from sparse datasets is presented. Instead of focussing on complete datasets and delivering results at the end of the computation process, our incremental approach delivers reasonable results while computing, for instance, to quickly visualize the virtual world or to create obstacle maps. 1
STATUTORY DECLARATION
, 2012
"... Ich erkläre an Eides statt, dass ich die vorliegende Arbeit selbstständig verfasst, andere als die angegebenen Quellen/Hilfsmittel nicht benutzt, und die den benutzten Quellen wörtlich und inhaltlich entnommene Stellen als solche kenntlich gemacht habe. ..."
Abstract
- Add to MetaCart
Ich erkläre an Eides statt, dass ich die vorliegende Arbeit selbstständig verfasst, andere als die angegebenen Quellen/Hilfsmittel nicht benutzt, und die den benutzten Quellen wörtlich und inhaltlich entnommene Stellen als solche kenntlich gemacht habe.
