Fast relocalization for visual odometry using binary features

Abstract

State-of-the-art visual odometry algorithms achieve remarkable efficiency and accuracy. Under realistic conditions, however, tracking failures are inevitable and to continue tracking, a recovery strategy is required. In this paper, we propose a relocalization system that enables realtime, 6D pose recovery for wide baselines. Our approach targets specifically resource-constrained hardware such as mobile phones. By exploiting the properties of low-complexity binary feature descriptors, nearest-neighbor search is performed efficiently using Locality Sensitive Hashing. Our method does not require time-consuming offline training of hash tables and it can be applied to any visual odometry system. We provide a thorough evaluation of effectiveness, robustness and runtime on an indoor test sequence with available ground truth poses. We investigate the system parameterization and compare the relocalization performance for the three binary descriptors BRIEF, unscaled BRIEF and ORB. In contrast to previous work on mobile visual odometry, we are able to quickly recover from tracking failures within maps with thousands of 3D feature points.

Indoor Pedestrian Localization With a Smartphone: A Comparison of Inertial and Vision-Based Methods

“…Thus an approach for wide-area 6DoF pose estimation relying on a previously acquired 3D feature model (image-to-map matching) was presented in [12]. Other authors [13] use binary features combined with Locality Sensitive Hashing (LSH) to estimate the pose of the camera within a Parallel Tracking and Mapping (PTAM) algorithm. For both methods, the obtained pose estimate is further refined using bundle adjustment approaches.…”

Low-latency Visual SLAM with Appearance-Enhanced Local Map Building

“…Hashing has been applied to modules of VSLAM where real-time performance is not required. [27] indexed binary descriptors with Locality Sensitive Hashing (LSH) [28], and demonstrated good relocalization performance in a VSLAM system. [29] utilized Multi-Index Hashing (MIH) [11] in the loop closing module of VSLAM.…”

Image-based camera localization: an overview

“…[Valentin et al 2015] trained a regression forest to predict mixtures of anisotropic 3D Gaussians and show how the predicted uncertainties can be taken into account for continuous pose optimization. [Straub et al 2013] From above the works of known environments, we see that fast camera localization from large data attract more and more attentions. This is because there are many applications for camera localization from large data, for example location based service, relocalization of SLAM for all kinds of robots, and AR navigations.…”