Image Based Localization in Urban Environments

By Jana Kosecka. 3D Data Processing Visualization and Transmission, 2006

427 citations recorded by the source corpus.

In this paper we present a prototype system for image based localization in urban environments. Given a database of views of city street scenes tagged by GPS locations, the system computes the GPS location of a novel query view. We first use a wide-baseline matching technique based on SIFT features to select the closest views in the database. Often due to a large change of viewpoint and presence of repetitive structures, a large percentage of matches (> 50%) are not correct correspondences. The subsequent motion estimation between the query view and the reference view, is then handled by a novel and efficient robust estimation technique capable of dealing with large percentage of outliers. This stage is also accompanied by a model selection step among the fundamental matrix and the homography. Once the motion between the closest reference views is estimated, the location of the query view is then obtained by triangulation of translation directions. Approximate solutions for cases when triangulation cannot be obtained reliably are also described. The presented system is tested on the dataset used in ICCV 2005 Computer Vision Contest and is shown to have higher accuracy than previous reported results.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/d4dd2f8852077682d3b7011f0806ac542ea64523

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

Visual simultaneous localization and mapping: a survey (2012, background): “Some details may be found in Tuytelaars and Van-Gool (2004); Zhang and Kosecka (2006) and Matas et al. (2002).”
Avoiding Confusing Features in Place Recognition (2010, background): “Most previous work on image-based place recognition focused on small scale settings [14,15,16].”
A review of volunteered geographic information quality assessment methods (2017, methodology): “On a related work, Zhang and Kosecka (2006) used feature-based geometric matching using the image recognition software SIFT (Lindeberg 2012) to localize sample photos in urban environments.”
Retrieving landmark and non-landmark images from community photo collections (2010, methodology): “Along with triangulation, it may also lead to exact localization as in Zhang and Kosecka [35].”
Efficient & Effective Prioritized Matching for Large-Scale Image-Based Localization (2017, methodology): “In contrast to our method, these approaches only approximate the pose of the query camera [14], [17], [18], [20].”
Clues from the beaten path: Location estimation with bursty sequences of tourist photos (2011, methodology): “The wide availability of online consumer photo collections in recent years has spurred much research in the field [14, 12, 21, 18, 15, 16, 4, 13, 6, 8, 7, 9].”
50 Years of object recognition: Directions forward (2013, background): “Local image-based features are also used for vision based localization and mapping with some success [165], [166], [167] and are currently quite popular in content-based image retrieval [168].”
Egocentric Shopping Cart Localization (2018, methodology): “The 6-DOF label of the retrieved image is hence attached to the query image in order to obtain the 6-DOF prediction [21]–[23].”
Fast image-based localization using direct 2D-to-3D matching (2011, methodology): “After finding those images in a database that are most similar to the query image, the location of the query can be determined relative to them [22, 31].”
Ultra-wide Baseline Facade Matching for Geo-localization (2012, positive): “Using [9] they match a query street-view image to a database of geo-tagged street-view images using wide-baseline matching.”
SKYLINE2GPS: Localization in urban canyons using omni-skylines (2010, methodology): “In the last few years, there has been an increasing interest in inferring geolocation from images without using SLAMstyle algorithms [25], [33], [13], [10].”
City-Scale Location Recognition (2007, background): “Vision-based location systems [10, 11, 14] have generally focused on small databases (200-1000 images) taken by human photographers.”
Image Retrieval for Image-Based Localization Revisited (2012, background): “Early works on image-based localization exhaustively compared a query image against all database images [27, 41].”
Location Recognition Using Prioritized Feature Matching (2010, background): “While there is a history of matching individual features for recognition and localization [13–15], we advocate reversing the usual process of matching.”
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.”
Accurate Image Localization Based on Google Maps Street View (2010, methodology): “[8] perform the localization task by matching image key points and then applying a geometrical alignment.”
Visual-inertial navigation, mapping and localization: A scalable real-time causal approach (2011, methodology): “…mapping and localization algorithms such as Zhu et al. (2007), Mei et al. (2009), Konolige and Agrawal (2008), Mourikis and Roumeliotis (2008), Zhang and Kosecka (2006), Wang et al. (2005), Williams et al. (2007), Goedeme et al. (2007), and Košecká et al. (2005), we exploit structural…”
Understanding the Limitations of CNN-Based Absolute Camera Pose Regression (2019, background): “More precise estimates can be obtained by using feature matches between the test image and the retrieved images for relative pose estimation [13,82,83].”