Long-Term Loop Closure Detection through Visual-Spatial Information Preserving Multi-Order Graph Matching

Gao, Peng; Zhang, Hao

doi:10.1609/aaai.v34i06.6604

Cited by 9 publications

(3 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We utilized both indoor and outdoor datasets to evaluate the performance of our methods, HGCN-FABMAP, HGCN-BoW, and HGCN-ORB. Our presented methods are benchmarked against FABMAP, BoW, ORB, LSD-SLAM, ORB-SLAM2, FOVIS, and methods described in references [20,21] and [22].…”

Section: Resultsmentioning

confidence: 99%

Semisupervised Vector Quantization in Visual SLAM Using HGCN

Zarringhalam,

Shiry Ghidary,

Mohades

et al. 2024

International Journal of Intelligent Systems

View full text Add to dashboard Cite

We present a novel vector quantization (VQ) module for the two state-of-the-art long-range simultaneous localization and mapping (SLAM) algorithms. The VQ task in SLAM is generally performed using unsupervised methods. We provide an alternative approach trough embedding a semisupervised hyperbolic graph convolutional neural network (HGCN) in the VQ step of the SLAM processes. The SLAM platforms we have utilized for this purpose are fast appearance-based mapping (FABMAP) and oriented fast and rotated short (ORB), both of which rely on extracting the features of the captured images in their loop closure detection (LCD) module. For the first time, we have considered the space formed by these SURF features, robust image descriptors, as a graph, enabling us to apply an HGCN in the VQ section which results in an improved LCD performance. The HGCN vector quantizes the SURF feature space, leading to a bag-of-word (BoW) representation construction of the images. This representation is subsequently used to determine LCD accuracy and recall. Our approaches in this study are referred to as HGCN-FABMAP and HGCN-ORB. The main advantage of using HGCN in the LCD section is that it scales linearly when the features are accumulated. The benchmarking experiments show the superiority of our methods in terms of both trajectory generation accuracy in small-scale paths and LCD accuracy and recall for large-scale problems.

show abstract

Section: Resultsmentioning

confidence: 99%

Semisupervised Vector Quantization in Visual SLAM Using HGCN

Zarringhalam,

Shiry Ghidary,

Mohades

et al. 2024

International Journal of Intelligent Systems

View full text Add to dashboard Cite

show abstract

“…Wang et al [50] introduced semantic topological graphs to encode the spatial information of landmarks and used random walk descriptors to characterize topological graphs for graph matching. Gao and Zhang [51] proposed a multi-order graph matching method for loop closure detection in addition to vector-based descriptors.…”

Section: Spatial-based Methodsmentioning

confidence: 99%

Loop Closure Detection Based on Differentiable Manifold

Dong

Xue

Zhang

et al. 2022

Mobile Information Systems

View full text Add to dashboard Cite

Loop closure detection is an important part of SLAM (simultaneous location and mapping), which can effectively reduce the cumulative error of the system after long period of exploration. The existing loop closure detection methods are mainly to evenly distribute the accumulated error in the robot trajectory, but the motion error of the actual robot is also related to its motion speed and rotation angle, while the corrected motion trajectory of the robot is difficult to match the real trajectory. Based on the analysis of the mechanism of robot motion error, this paper proposes a novel loop closure detection method based on differentiable manifold, which mainly includes real-time pose based on manifold tangent space and smooth motion trajectory model of robot based on differential geometry. Firstly, we introduce the Frenet framework structure and establish the corresponding manifold tangent space theory for the keyframe pose nodes. The real-time problem of robot motion is equivalent to the problem of finding the optimal angle tangent vector. Secondly, the motion speed between keyframes is used to determine the characteristics of the robot motion trajectory. We calculate the curvature and torsion of the curve composed of several nodes based on the manifold tangent space and then combine the curve interpolation and ﬁtting of the keyframe nodes to achieve the approximation of the robot motion trajectory, and the smooth curve of the robot trajectory is obtained. Finally, the experiment veriﬁes that the method in this paper can effectively ensure the continuity and smoothness of the robot’s trajectory, thereby reducing the cumulative error of the system and improving the accuracy of loop closure detection.

show abstract

“…To overcome this, some approaches utilized image sequence matching (Milford and Wyeth 2012;Hansen et al 2014;Naseer et al 2014;Doan et al 2019;Lu et al 2021;Garg and Milford 2021) to achieve robust VPR under extreme variations in illumination, weather, and season. And other methods (Sünderhauf et al 2015;Chen et al 2017b;Xin et al 2019;Gao et al 2020) mined discriminative landmarks for VPR. Moreover, VPR models commonly require training on large-scale place datasets with weak supervision.…”

Section: Related Workmentioning

confidence: 99%

Deep Homography Estimation for Visual Place Recognition

Lu,

Dong,

Zhang

et al. 2024

AAAI

View full text Add to dashboard Cite

Visual place recognition (VPR) is a fundamental task for many applications such as robot localization and augmented reality. Recently, the hierarchical VPR methods have received considerable attention due to the trade-off between accuracy and efficiency. They usually first use global features to retrieve the candidate images, then verify the spatial consistency of matched local features for re-ranking. However, the latter typically relies on the RANSAC algorithm for fitting homography, which is time-consuming and non-differentiable. This makes existing methods compromise to train the network only in global feature extraction. Here, we propose a transformer-based deep homography estimation (DHE) network that takes the dense feature map extracted by a backbone network as input and fits homography for fast and learnable geometric verification. Moreover, we design a re-projection error of inliers loss to train the DHE network without additional homography labels, which can also be jointly trained with the backbone network to help it extract the features that are more suitable for local matching. Extensive experiments on benchmark datasets show that our method can outperform several state-of-the-art methods. And it is more than one order of magnitude faster than the mainstream hierarchical VPR methods using RANSAC. The code is released at https://github.com/Lu-Feng/DHE-VPR.

show abstract

Long-Term Loop Closure Detection through Visual-Spatial Information Preserving Multi-Order Graph Matching

Cited by 9 publications

References 31 publications

Semisupervised Vector Quantization in Visual SLAM Using HGCN

Semisupervised Vector Quantization in Visual SLAM Using HGCN

Loop Closure Detection Based on Differentiable Manifold

Deep Homography Estimation for Visual Place Recognition

Contact Info

Product

Resources

About