Altitude-Aware Visual Place Recognition in Top-Down View

TL;DR

This study proposes an altitude-adaptive VPR approach that integrates ground feature density analysis with image classification techniques, offering a practical and scalable solution for accurate airborne platforms localization under large altitude variations and limited sensor availability.

Abstract

To address the challenge of aerial visual place recognition (VPR) problem under significant altitude variations, this study proposes an altitude-adaptive VPR approach that integrates ground feature density analysis with image classification techniques. The proposed method estimates airborne platforms' relative altitude by analyzing the density of ground features in images, then applies relative altitude-based cropping to generate canonical query images, which are subsequently used in a classification-based VPR strategy for localization. Extensive experiments across diverse terrains and altitude conditions demonstrate that the proposed approach achieves high accuracy and robustness in both altitude estimation and VPR under significant altitude changes. Compared to conventional methods relying on barometric altimeters or Time-of-Flight (ToF) sensors, this solution requires no additional hardware and offers a plug-and-play solution for downstream applications, {making it suitable for small- and medium-sized airborne platforms operating in diverse environments, including rural and urban areas.} Under significant altitude variations, incorporating our relative altitude estimation module into the VPR retrieval pipeline boosts average R@1 and R@5 by 29.85\% and 60.20\%, respectively, compared with applying VPR retrieval alone. Furthermore, compared to traditional {Monocular Metric Depth Estimation (MMDE) methods}, the proposed method reduces the mean error by 202.1 m, yielding average additional improvements of 31.4\% in R@1 and 44\% in R@5. These results demonstrate that our method establishes a robust, vision-only framework for three-dimensional visual place recognition, offering a practical and scalable solution for accurate airborne platforms localization under large altitude variations and limited sensor availability.

Figures (5)

• Figure 1: Overall pipeline for relative altitude estimation and visual place recognition (VPR). The pipeline proceeds sequentially from steps ① to ④, where steps ① and ② are offline preparation stages and steps ③ and ④ are online inference stages.
• Figure 2: *Schematic diagram of the processing procedure of the sample query images from 4 datasets. Among them, the frequency-domain $I^{\text{freq}}_{\text{in}}$ have undergone brightness transformation for visual presentation effect.
• Figure 3: The process to transform an input image to a primitive image.
• Figure 4: VPR of the primitive images.
• Figure 5: Relative Altitude Estimation Results