Building Location Models for Visual Place Recognition

Stumm, Elena; Mei, Christopher; Lacroix, Simon

doi:10.1177/0278364915570140

Cited by 27 publications

(25 citation statements)

References 33 publications

(120 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The use of this representation permits the analogous application of many theoretical developments such as tf-idf (term frequency×inverse document frequency) and probabilistic naive-bayes [9], from the fields of text retrieval and classification, on images [7], [3]. Such techniques apply well to place recognition for mobile robots, and are generally well established in the field, including extended generative models for location observations [10], [11], [6].…”

Section: A Visual Feature Based Place Recognitionmentioning

confidence: 99%

“…Alternatively, instead of using additional distance and position measurements, in [6], the implicit geometric relationships between features are given by covisibility information. Landmarks are tracked between successive images using a single camera, recording the covisibility between landmarks in a graph-based map of the world.…”

Section: A Visual Feature Based Place Recognitionmentioning

confidence: 99%

“…Such a covisibility map has also been successfully used in [6] to dynamically extract and retrieve locations as clusters from the graph, which provide better matches to a query location than other methods relying on locations represented by fixed image frames. Building on this work, here we present a novel approach to efficiently incorporate more comprehensive structural comparisons between candidate locations, illustrating improvement in a variety of test cases as shown in Section IV.…”

Section: A Visual Feature Based Place Recognitionmentioning

confidence: 99%

“…Building on previous work in [6], we use the covisibility map of landmarks and their associated visual words to provide subgraphs, which serve as locations. Sections III-A and III-B provide a summary of this framework for completeness, followed by a description of novel location representations and comparison techniques in the remainder of this section.…”

Section: Location Graphsmentioning

confidence: 99%

“…Considering a location initially as a constellation of visual landmarks, a graph is constructed, such that nodes are labeled with the associated visual words and edges correspond to binary covisibility of these words, i.e. connecting two nodes if the underlying landmarks have been co-observed in an image, as in [5], [6]. These location graphs reside as subgraphs in a larger covisibility map spanning the explored environment.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Location graphs for visual place recognition

Stumm

Mei

Lacroix

et al. 2015

2015 IEEE International Conference on Robotics and Automation (ICRA)

Self Cite

View full text Add to dashboard Cite

Abstract-With the growing demand for deployment of robots in real scenarios, robustness in the perception capabilities for navigation lies at the forefront of research interest, as this forms the backbone of robotic autonomy. Existing place recognition approaches traditionally follow the feature-based bag-of-words paradigm in order to cut down on the richness of information in images. As structural information is typically ignored, such methods suffer from perceptual aliasing and reduced recall, due to the ambiguity of observations. In a bid to boost the robustness of appearance-based place recognition, we consider the world as a continuous constellation of visual words, while keeping track of their covisibility in a graph structure. Locations are queried based on their appearance, and modelled by their corresponding cluster of landmarks from the global covisibility graph, which retains important relational information about landmarks. Complexity is reduced by comparing locations by their graphs of visual words in a simplified manner. Test results show increased recall performance and robustness to noisy observations, compared to state-of-the-art methods.

show abstract