From structure-from-motion point clouds to fast location recognition

By Arnold Irschara, Christopher Zach, Jan-Michael Frahm, Horst Bischof. 2009

593 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Aerospace Engineering

In Computer Vision and Pattern Recognition

Citations in the corpus

CPlaNet: Enhancing Image Geolocalization by Combinatorial Partitioning of Maps (2018, methodology): “Pose estimation approaches [9,18,19,20,21,22,23] match query images against 3D models of an area, and employ 2D-3D feature correspondences to identify 6-DOF query poses.”
PlaNet - Photo Geolocation with Convolutional Neural Networks (2016, methodology): “Because matching the query descriptors against the 3D model descriptors is expensive, some approaches combine this technique with efﬁ-cient image retrieval based on inverted indices [5, 21, 42].”
Understanding the Limitations of CNN-Based Absolute Camera Pose Regression (2019, methodology): “Image retrieval has also been used as part of structure-based approaches [14,26,57].”
Back to the Feature: Learning Robust Camera Localization from Pixels to Pose (2021, background): “Such approaches detect, describe [7, 49], and match [32, 46, 48, 73, 81, 96] local features, maintain an explicit sparse 3D representation of the environment, and sometimes leverage image retrieval [33, 86] to scale to large scenes [32, 57, 70, 75, 82, 87].”
Efficient Global 2D-3D Matching for Camera Localization in a Large-Scale 3D Map (2017, methodology): “[23] first proposed the 2D-3D matching method for camera localization.”
Predicting Good Features for Image Geo-Localization Using Per-Bundle VLAD (2015, methodology): “The other is to estimate the full camera pose of the query image using a 3D structure-from-motion model constructed from reference images [13, 17, 25, 31], which is limited to places with a dense distribution of reference images.”
Location Recognition Using Prioritized Feature Matching (2010, methodology): “Note that this is different from the covering problem in [4], which aims to cover the points instead of the images.”
Benchmarking 6DOF Outdoor Visual Localization in Changing Conditions (2018, background): “52M (69) 15,179 1000 San Francisco [16, 34, 57] Urban Free Viewpoint 30M / 149M (1) 610,773 442 X(SfM) Vienna [29] Landmarks Free Viewpoint 1.”
Fast image-based localization using direct 2D-to-3D matching (2011, methodology): “We evaluate all methods on the three datasets presented in [13, 16], kindly provided by Li et al.”
Privacy Preserving Image Queries for Camera Localization (2019, methodology): “Structure-based camera pose estimation methods are now quite mature [28,38,40,54,79] and deployed in many products (e.”
Worldwide Pose Estimation Using 3D Point Clouds (2012, methodology): “[22] use SfM reconstructions to generate a set of “virtual” images that cover a scene, then index these as documents using BoW methods.”
From single image query to detailed 3D reconstruction (2015, background): “For large zoom factors, it is not enough to simply assume a default focal length [10, 26], inferred from the image dimensions, and use it as an initial estimate for a nonlinear refinement.”
24/7 Place Recognition by View Synthesis (2018, unknown): “[43] generate bag-of-visual-word descriptors extracted from existing views for virtual locations on a map to better model scene visibility.”
Structure-from-Motion Revisited (2016, background): “More visible points and a more uniform distribution of these points should result in a higher score S [31], such that images with a better-conditioned configuration of visible points are registered first.”
Visual Place Recognition with Repetitive Structures (2015, background): “Examples include: (i) learning better visual vocabularies [21, 28]; (ii) developing quantization methods less prone to quantization errors [14, 27, 44]; (iii) combining returns from multiple query images depicting the same scene [4, 6]; (iv) exploiting the 3D or graph structure of the database [11, ”
Learned Contextual Feature Reweighting for Image Geo-Localization (2017, methodology): “Image geo-localization has been cast as an image retrieval task [2, 10, 19, 28, 66], a 2D to 3D registration task [17, 21, 32, 47, 46], and a per-location classification task [16, 62].”
Improving Image-Based Localization by Active Correspondence Search (2012, methodology): “Images from the Panoramio website are used for testing [1].”
From Coarse to Fine: Robust Hierarchical Localization at Large Scale (2018, background): “Hierarchical localization [19, 30, 40] takes a different approach by dividing the problem into a global, coarse search followed by a fine pose estimation.”
Handling Urban Location Recognition as a 2D Homothetic Problem (2010, background): “Location recognition approaches [9,8,6] usually know the intrinsic parameters of the camera, but do not exploit dense 3D models of the scene since these are diﬃcult to obtain for larger environments.”