Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Enhancing the Quality of 3D Lunar Maps Using JAXA's Kaguya Imagery

Yumi Iwashita, Haakon Moe, Yang Cheng, Adnan Ansar, Georgios Georgakis, Adrian Stoica, Kazuto Nakashima, Ryo Kurazume, Jim Torresen

TL;DR

This paper tackles elevation errors in 3D lunar maps generated from JAXA's Kaguya Terrain Camera (TC) imagery caused by JPEG compression artifacts. It analyzes how compression introduces disparity noise, especially in darker regions, and proposes a residual-learning approach using Palette (conditional diffusion) and IGEV++ to refine disparity maps; training centers on predicting the residual between compressed and uncompressed disparities. Experiments on ~70 stereo pairs show Palette provides the strongest reduction in disparity noise, achieving a mean residual of ~0.007 px and a standard deviation of ~0.033 px, corresponding to roughly 0.54 m elevation noise, a substantial improvement over the ~1.26 m noise from uncorrected compression. The work demonstrates the viability of enhancing 3D lunar maps from legacy compressed data, potentially improving rover navigation safety for missions like Endurance, and outlines directions to broaden evaluation across more DN values and JPEG tables.

Abstract

As global efforts to explore the Moon intensify, the need for high-quality 3D lunar maps becomes increasingly critical-particularly for long-distance missions such as NASA's Endurance mission concept, in which a rover aims to traverse 2,000 km across the South Pole-Aitken basin. Kaguya TC (Terrain Camera) images, though globally available at 10 m/pixel, suffer from altitude inaccuracies caused by stereo matching errors and JPEG-based compression artifacts. This paper presents a method to improve the quality of 3D maps generated from Kaguya TC images, focusing on mitigating the effects of compression-induced noise in disparity maps. We analyze the compression behavior of Kaguya TC imagery, and identify systematic disparity noise patterns, especially in darker regions. In this paper, we propose an approach to enhance 3D map quality by reducing residual noise in disparity images derived from compressed images. Our experimental results show that the proposed approach effectively reduces elevation noise, enhancing the safety and reliability of terrain data for future lunar missions.

Enhancing the Quality of 3D Lunar Maps Using JAXA's Kaguya Imagery

TL;DR

This paper tackles elevation errors in 3D lunar maps generated from JAXA's Kaguya Terrain Camera (TC) imagery caused by JPEG compression artifacts. It analyzes how compression introduces disparity noise, especially in darker regions, and proposes a residual-learning approach using Palette (conditional diffusion) and IGEV++ to refine disparity maps; training centers on predicting the residual between compressed and uncompressed disparities. Experiments on ~70 stereo pairs show Palette provides the strongest reduction in disparity noise, achieving a mean residual of ~0.007 px and a standard deviation of ~0.033 px, corresponding to roughly 0.54 m elevation noise, a substantial improvement over the ~1.26 m noise from uncorrected compression. The work demonstrates the viability of enhancing 3D lunar maps from legacy compressed data, potentially improving rover navigation safety for missions like Endurance, and outlines directions to broaden evaluation across more DN values and JPEG tables.

Abstract

As global efforts to explore the Moon intensify, the need for high-quality 3D lunar maps becomes increasingly critical-particularly for long-distance missions such as NASA's Endurance mission concept, in which a rover aims to traverse 2,000 km across the South Pole-Aitken basin. Kaguya TC (Terrain Camera) images, though globally available at 10 m/pixel, suffer from altitude inaccuracies caused by stereo matching errors and JPEG-based compression artifacts. This paper presents a method to improve the quality of 3D maps generated from Kaguya TC images, focusing on mitigating the effects of compression-induced noise in disparity maps. We analyze the compression behavior of Kaguya TC imagery, and identify systematic disparity noise patterns, especially in darker regions. In this paper, we propose an approach to enhance 3D map quality by reducing residual noise in disparity images derived from compressed images. Our experimental results show that the proposed approach effectively reduces elevation noise, enhancing the safety and reliability of terrain data for future lunar missions.

Paper Structure

This paper contains 7 sections, 5 figures, 4 tables.

Table of Contents

  1. INTRODUCTION
  2. JPEG COMPRESSION ON KAGUYA TC IMAGES AND IMPROVING DISPARITY IMAGES
  3. Kaguya TC images and JPEG compression
  4. Improve disparity images affected by compression-induced errors
  5. EXPERIMENTS
  6. Comparison of Palette, IGEV++, and low pass filter
  7. CONCLUSIONS AND FUTURE WORK

Figures (5)

  • Figure 1: Illustrative example of elevation error ( 20 m) in a 3D reconstruction derived from compressed stereo imagery (left). The DEM generated from uncompressed imagery of the same area (right) appears smooth.
  • Figure 2: An example stereo pair (left: TC1W2B0$\_$01$\_$01685N188E0047, right: TC2W2B0$\_$01$\_$01685N185E0048).
  • Figure 3: Illustrative comparison of disparity values from compressed (orange) and uncompressed (blue) images.
  • Figure 4: Example visualizations from the test dataset include: (a) TC1 images from three example rectified stereo pairs (Stereo pair IDs: 166, 615, and 918 from left to right); (b) ground truth disparity residuals corresponding to the stereo pairs in (a); and (c–e) estimated disparity residuals produced by Palette, IGEV++, and LPF
  • Figure 5: An example visualizations of 2D plots with the x-axis representing predictions and the y-axis representing ground truth (Stereo pair ID = 166).