Handling Urban Location Recognition as a 2D Homothetic Problem

Baatz, Georges; Köser, Kevin; Chen, David; Grzeszczuk, Radek; Pollefeys, Marc

doi:10.1007/978-3-642-15567-3_20

Cited by 61 publications

(43 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Further information, such as building outlines or 3D geometry, can be combined with the image database to improve localization approaches [2,4]. When the coverage of the scene by the images is dense enough, SfM techniques can be used to reconstruct a 3D point cloud of the scene [8,10,31,33].…”

Section: Introductionmentioning

confidence: 99%

Image Retrieval for Image-Based Localization Revisited

Sattler

Weyand

Leibe

et al. 2012

Procedings of the British Machine Vision Conference 2012

276

233

View full text Add to dashboard Cite

To reliably determine the camera pose of an image relative to a 3D point cloud of a scene, correspondences between 2D features and 3D points are needed. Recent work has demonstrated that directly matching the features against the points outperforms methods that take an intermediate image retrieval step in terms of the number of images that can be localized successfully. Yet, direct matching is inherently less scalable than retrievalbased approaches. In this paper, we therefore analyze the algorithmic factors that cause the performance gap and identify false positive votes as the main source of the gap. Based on a detailed experimental evaluation, we show that retrieval methods using a selective voting scheme are able to outperform state-of-the-art direct matching methods. We explore how both selective voting and correspondence computation can be accelerated by using a Hamming embedding of feature descriptors. Furthermore, we introduce a new dataset with challenging query images for the evaluation of image-based localization.

show abstract

Section: Introductionmentioning

confidence: 99%

Image Retrieval for Image-Based Localization Revisited

Sattler

Weyand

Leibe

et al. 2012

Procedings of the British Machine Vision Conference 2012

276

233

View full text Add to dashboard Cite

show abstract

“…Orthorectification, essentially facilitates the alignment of images taken from different viewpoints to form a larger mosaic, and as shown in Baatz et al [2] the overlapping part of two orthophotos of the same place is typically very similar resulting to their straightforward alignment. Testing on imagery of buildings facades, [2] factorize the rotation out of the recognition problem by generating gravity-aligned orthophotos outperforming purely 2D-based methods. In a similar spirit, Chen et al [4] demonstrate a gain in place recognition by combining both unmodified perspective images and their corresponding orthophotos.…”

Section: Related Workmentioning

confidence: 99%

“…Inspired by [2], in this paper, instead of searching for a perfect alignment between images, we aim to verify whether the configuration of features shared by two orthophotos presents a consistent layout. This step is known as a geometrical check in loop-closure algorithms.…”

Section: Related Workmentioning

confidence: 99%

“…This step is known as a geometrical check in loop-closure algorithms. Moreover, in this work, the orthophoto plane is directly extracted from the 3D landmarks used for the robot visual navigation system without the need of computing lines and extracting vanishing points, as in [2].…”

Section: Related Workmentioning

confidence: 99%

“…A geometric check based on a RANSAC scheme is usually applied after appearance matching to improve loop closure detection by verifying whether the configuration of features belonging to these two images presents a consistent layout. When matching gravity-compatible orthophotos, Baatz et al [2] reduce the 6 DOF perspective recognition problem to a homothetic problem involving only scale and a translation in a 2D plane. By exploiting the fact that they are solving a homothetic problem, [2] suggests to replace the computationally expensive RANSAC-based geometric check by an efficient 1D voting scheme, where scale as well as a horizontal and a vertical offset are estimated separately.…”

Section: Geometric Checkmentioning

confidence: 99%

See 2 more Smart Citations

Loop-Closure Detection in Urban Scenes for Autonomous Robot Navigation

Maffra

Teixeira

Chen

et al. 2017

2017 International Conference on 3D Vision (3DV)

View full text Add to dashboard Cite

Relocalization is a vital process for autonomous robot navigation, typically running in the background of sequential localization and mapping to detect loops in the robot's trajectory. Such loop-closure detections enable corrections for drift accumulated during the estimation processes and even recovery from complete localization failures. In this work, we present a novel approach loosely integrated with a keyframe-based SLAM system to perform loop-closure detection in urban scenarios for autonomous robot navigation. Generating a mesh of the current robot's surroundings in real-time using monocular and inertial cues, the proposed method estimates the most salient plane in the current view, enabling the creation of the corresponding orthophoto for this plane. Evaluating image similarity on orthophotos forms a much better conditioned problem for relocalization, minimizing effects from viewpoint changes. Employing binary image descriptors and tests on their relative constellation in the image, the proposed approach exhibits robustness also to illumination and situational variations common in real scenes, overall resulting to significant improvement in loop-closure detection performance in urban scenes with respect to the state of the art.Video -https://www.youtube.com/c/V4RLteam

show abstract

CPlaNet: Enhancing Image Geolocalization by Combinatorial Partitioning of Maps

Seo

Weyand

Sim

et al. 2018

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Image geolocalization is the task of identifying the location depicted in a photo based only on its visual information. This task is inherently challenging since many photos have only few, possibly ambiguous cues to their geolocation. Recent work has cast this task as a classification problem by partitioning the earth into a set of discrete cells that correspond to geographic regions. The granularity of this partitioning presents a critical trade-off; using fewer but larger cells results in lower location accuracy while using more but smaller cells reduces the number of training examples per class and increases model size, making the model prone to overfitting. To tackle this issue, we propose a simple but effective algorithm, combinatorial partitioning, which generates a large number of fine-grained output classes by intersecting multiple coarse-grained partitionings of the earth. Each classifier votes for the fine-grained classes that overlap with their respective coarse-grained ones. This technique allows us to predict locations at a fine scale while maintaining sufficient training examples per class. Our algorithm achieves the state-of-the-art performance in location recognition on multiple benchmark datasets.

show abstract

Handling Urban Location Recognition as a 2D Homothetic Problem

Cited by 61 publications

References 14 publications

Image Retrieval for Image-Based Localization Revisited

Image Retrieval for Image-Based Localization Revisited

Loop-Closure Detection in Urban Scenes for Autonomous Robot Navigation

CPlaNet: Enhancing Image Geolocalization by Combinatorial Partitioning of Maps

Contact Info

Product

Resources

About