pith. sign in

arxiv: 2604.01032 · v3 · submitted 2026-04-01 · 💻 cs.CV

Sub-metre Lunar DEM Generation and Validation from Chandrayaan-2 OHRC Multi-View Imagery Using an Open-Source Pipeline

Aaranay Aadi , Jai Singla , Nitant Dube , Oleg Alexandrov This is my paper

Pith reviewed 2026-05-13 22:31 UTC · model grok-4.3

classification 💻 cs.CV
keywords lunar DEMChandrayaan-2 OHRCstereo photogrammetryopen-source pipelinesub-metre resolutionNAC validationlunar topography
0
0 comments X

The pith

Sub-metre lunar DEMs are generated from Chandrayaan-2 OHRC multi-view imagery via an open-source pipeline and validated to 5.85 m vertical RMSE.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper shows how to turn non-paired high-resolution images from Chandrayaan-2's Orbiter High Resolution Camera into digital elevation models at 24 to 54 cm spacing. It selects usable stereo pairs by computing baseline-to-height ratios and convergence angles from image metadata, then applies dense matching and ray triangulation to build point clouds that are gridded into DEMs. These models are aligned to existing NAC terrain references using ICP and a constant offset removal, producing vertical RMSE of 5.85 m and horizontal accuracy within one pixel. Readers would care because such local topography supports safer landing site selection and rover path planning where coarser global models leave gaps. The work demonstrates that free tools can extract usable detail from new orbital data without proprietary processing.

Core claim

The paper establishes that an open-source photogrammetric pipeline applied to OHRC multi-view imagery produces sub-metre DEMs at effective resolutions of 24-54 cm across five lunar sites; after ICP alignment to NAC DTMs and constant-bias correction the resulting surfaces achieve 5.85 m vertical RMSE and horizontal accuracy within approximately 30 cm.

What carries the argument

Geometric selection of stereo pairs via B/H ratio and convergence angle computation, followed by dense stereo correspondence, ray triangulation to point clouds, gridding, and ICP alignment plus constant-bias correction to NAC references.

Load-bearing premise

The chosen OHRC stereo pairs have enough geometric quality that ICP alignment and a single constant-bias offset can remove all systematic errors without creating new distortions in the relative elevations.

What would settle it

Independent laser altimeter profiles or withheld ground-control points measured at the same sites showing vertical differences larger than 5.85 m RMSE after the described alignment steps would falsify the accuracy result.

Figures

Figures reproduced from arXiv: 2604.01032 by Aaranay Aadi, Jai Singla, Nitant Dube, Oleg Alexandrov.

Figure 1
Figure 1. Figure 1: Overview of the OHRC DEM generation pipeline. Stages proceed from multi-view OHRC imagery and stereo [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: DEMs for four of the reconstructed regions. Region 5 is demonstrated in section 7 for its high void fraction. [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Visual feature comparison of OHRC generated DEM vs NAC DTM (both hillshaded) for Sites 1 and 2, [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Visual feature comparison of OHRC generated DEM vs NAC DTM (both hillshaded) for Sites 3 and 4, [PITH_FULL_IMAGE:figures/full_fig_p010_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Terrain profile comparison across Region 4. Horizontal axis: distance along profile transect (m). Vertical axis: [PITH_FULL_IMAGE:figures/full_fig_p012_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Terrain profile comparison across Region 1. Horizontal axis: distance along profile transect (m). Vertical axis: [PITH_FULL_IMAGE:figures/full_fig_p012_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Terrain profile comparison across Region 2. Horizontal axis: distance along profile transect (m). Vertical axis: [PITH_FULL_IMAGE:figures/full_fig_p013_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Terrain profile comparison across a small patch of Region 2. Horizontal axis: distance along profile transect [PITH_FULL_IMAGE:figures/full_fig_p013_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Region 5 demonstrated with high void fraction [PITH_FULL_IMAGE:figures/full_fig_p015_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Region-of-interest stereo reconstruction. Left and centre: the left-camera and right-camera cropped image [PITH_FULL_IMAGE:figures/full_fig_p016_10.png] view at source ↗
read the original abstract

High-resolution digital elevation models (DEMs) of the lunar surface are essential for surface mobility planning, landing site characterization, and planetary science. The Orbiter High Resolution Camera (OHRC) on board Chandrayaan-2 has the best ground sampling capabilities of any lunar orbital imaging currently in use by acquiring panchromatic imagery at a resolution of roughly 20-30 cm per pixel. This work presents, for the first time, the generation of sub-metre DEMs from OHRC multi-view imagery using an exclusively open-source pipeline. Candidate stereo pairs are identified from non-paired OHRC archives through geometric analysis of image metadata, employing baseline-to-height (B/H) ratio computation and convergence angle estimation. Dense stereo correspondence and ray triangulation are then applied to generate point clouds, which are gridded into DEMs at effective spatial resolutions between approximately 24 and 54 cm across five geographically distributed lunar sites. Absolute elevation consistency is established through Iterative Closest Point (ICP) alignment against Lunar Reconnaissance Orbiter Narrow Angle Camera (NAC) Digital Terrain Models, followed by constant-bias offset correction. Validation against NAC reference terrain yields a vertical RMSE of 5.85 m (at native OHRC resolution), and a horizontal accuracy of within a pixel (approximately 30 cm) assessed by planimetric feature matching.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript describes an open-source pipeline to generate sub-metre DEMs from Chandrayaan-2 OHRC multi-view panchromatic imagery. Stereo pairs are selected via B/H ratio and convergence angle analysis, dense matching and ray triangulation produce point clouds that are gridded into DEMs at 24-54 cm effective resolution over five lunar sites. These DEMs are aligned to NAC reference DTMs using ICP followed by constant-bias correction, yielding a reported vertical RMSE of 5.85 m and horizontal accuracy within one pixel (~30 cm) via planimetric matching.

Significance. If the accuracy claims are substantiated, the work is significant for providing the first demonstration of sub-metre lunar DEM generation from OHRC data using fully open-source methods. This enables wider community access for applications in landing site characterization, rover navigation planning, and geological analysis. The use of multiple geographically distributed sites and external NAC validation strengthens the practical utility, though the independence of the reported error metrics from the alignment procedure requires explicit demonstration.

major comments (2)
  1. [Abstract / Validation] Abstract and validation section: The vertical RMSE of 5.85 m is reported after ICP alignment and constant-bias offset correction to NAC DTMs. Because the alignment explicitly minimizes discrepancies to the reference, this value primarily quantifies residual stereo-matching noise and local distortions rather than the absolute geopositioning accuracy of the OHRC-derived DEM. Pre-alignment error statistics or validation against an independent dataset (such as LOLA laser altimetry) should be included to support the absolute accuracy claim.
  2. [Methods (pair selection)] Methods (pair selection): The B/H ratio threshold and convergence angle criteria used to select candidate stereo pairs are described qualitatively but not quantified on a per-site basis. Without these values and an analysis of their impact on the resulting DEM quality, it is difficult to assess whether the input geometry consistently supports the claimed sub-metre horizontal performance across all five sites.
minor comments (2)
  1. [Abstract] Clarify how the 'effective spatial resolutions between approximately 24 and 54 cm' are computed from the native 20-30 cm GSD imagery and the gridding process.
  2. [Validation] Provide details on the full error propagation through the pipeline, including sensitivity to pair selection criteria.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive feedback on our manuscript. We have carefully considered each major comment and provide point-by-point responses below. We plan to make revisions to address the concerns raised.

read point-by-point responses

  1. Referee: Abstract and validation section: The vertical RMSE of 5.85 m is reported after ICP alignment and constant-bias offset correction to NAC DTMs. Because the alignment explicitly minimizes discrepancies to the reference, this value primarily quantifies residual stereo-matching noise and local distortions rather than the absolute geopositioning accuracy of the OHRC-derived DEM. Pre-alignment error statistics or validation against an independent dataset (such as LOLA laser altimetry) should be included to support the absolute accuracy claim.

    Authors: We agree with the referee that the reported vertical RMSE of 5.85 m is calculated after performing ICP alignment and constant-bias correction, which removes systematic translational differences between the OHRC DEM and the NAC reference. Consequently, this metric primarily reflects the residual errors due to stereo matching inaccuracies and local distortions rather than the absolute geopositioning accuracy. To address this, we will revise the manuscript to explicitly clarify this distinction and include pre-alignment error statistics (e.g., initial RMSE and mean bias before correction) in the validation section. While validation against LOLA altimetry would provide an independent absolute reference, it is beyond the scope of the current work due to the sparse nature of LOLA data; however, we will add a discussion on this limitation and the suitability of NAC for relative validation. revision: yes

  2. Referee: Methods (pair selection): The B/H ratio threshold and convergence angle criteria used to select candidate stereo pairs are described qualitatively but not quantified on a per-site basis. Without these values and an analysis of their impact on the resulting DEM quality, it is difficult to assess whether the input geometry consistently supports the claimed sub-metre horizontal performance across all five sites.

    Authors: We thank the referee for pointing this out. The manuscript describes the B/H ratio and convergence angle criteria qualitatively to explain the pair selection process. In the revised version, we will add a supplementary table or section that quantifies these values for each stereo pair at the five sites. Additionally, we will include an analysis examining the relationship between these geometric parameters and the resulting DEM quality indicators, such as point cloud density and achieved horizontal accuracy, to demonstrate that the selected geometries consistently support the sub-metre performance. revision: yes

Circularity Check

0 steps flagged

Validation uses external NAC DTMs after ICP alignment; no internal reduction of accuracies to fitted parameters from the same data

full rationale

The paper presents an open-source pipeline: geometric selection of OHRC stereo pairs via B/H ratio and convergence angle, dense stereo correspondence, ray triangulation to point clouds, and gridding to DEMs at 24-54 cm resolution. Absolute consistency is obtained by ICP alignment to external NAC DTMs plus constant-bias correction, after which validation reports vertical RMSE 5.85 m and horizontal accuracy within one pixel. No equations in the described chain reduce the reported accuracies to parameters fitted from the target OHRC data itself; the alignment reference is independent. No self-citations are load-bearing for the central claims, and no ansatz or uniqueness theorem is smuggled in. This yields a minor score reflecting reliance on external benchmarks rather than any circular reduction by construction.

Axiom & Free-Parameter Ledger

2 free parameters · 1 axioms · 0 invented entities

The pipeline assumes standard camera models and ephemeris accuracy from the OHRC instrument; no new physical constants or entities are introduced. Free parameters include B/H ratio thresholds and convergence angle cutoffs used for pair selection, plus grid resolution choices.

free parameters (2)
  • B/H ratio threshold
    Used to filter candidate stereo pairs; value not specified in abstract but directly affects which images are processed.
  • DEM grid resolution
    Chosen between 24-54 cm; affects final output spacing and is selected per site.
axioms (1)
  • domain assumption OHRC camera model and spacecraft ephemeris are sufficiently accurate for ray triangulation
    Invoked when performing dense stereo correspondence and point cloud generation.

pith-pipeline@v0.9.0 · 5553 in / 1233 out tokens · 49184 ms · 2026-05-13T22:31:35.974447+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. DEM Refinement and Validation on the Lunar Surface Using Shape-from-Shading with Chandrayaan-2 OHRC Imagery

    cs.CV 2026-04 unverdicted novelty 5.0

    A shape-from-shading framework refines sub-meter lunar DEMs with independent OHRC images at three sites, showing slope statistic improvements and identifying limits from viewing geometry.

Reference graph

Works this paper leans on

14 extracted references · 14 canonical work pages · cited by 1 Pith paper

  1. [1]

    doi:10.1016/0032-0633(95)00107-7. S. Agarwal, K. Mierle, and The Ceres Solver Team. Ceres solver,

    work page doi:10.1016/0032-0633(95)00107-7

  2. [2]

    MK Barker, E Mazarico, GA Neumann, MT Zuber, Junichi Haruyama, and DE Smith

    doi:10.1126/science.1164146. MK Barker, E Mazarico, GA Neumann, MT Zuber, Junichi Haruyama, and DE Smith. A new lunar digital elevation model from the lunar orbiter laser altimeter and selene terrain camera.Icarus, 273:346–355,

    work page doi:10.1126/science.1164146

  3. [3]

    doi:10.1109/34.121791. R. A. Beyer, O. Alexandrov, and S. McMichael. The Ames Stereo Pipeline: NASA’s open-source software for deriving and processing terrain data.Earth and Space Science, 5(9):537–548,

    work page doi:10.1109/34.121791

  4. [4]

    doi:10.1029/2018EA000409. A. R. Chowdhury, M. Saxena, A. Kumar, S. R. Joshi, A. Dagar, M. Mittal, S. Kirkire, J. Desai, D. Shah, J. C. Karelia, and A. Kumar. Orbiter high resolution camera onboard Chandrayaan-2 orbiter.Current Science, 118(4):560–565,

    work page doi:10.1029/2018ea000409

  5. [5]

    Community Sensor Model Working Group

    doi:10.18520/cs/v118/i4/560-565. Community Sensor Model Working Group. Community sensor model technical requirements document. Technical Report NGA.STND.0017 3.0, National Geospatial-Intelligence Agency, Springfield, V A,

    work page doi:10.18520/cs/v118/i4/560-565

  6. [6]

    Gabriele Facciolo, Carlo De Franchis, and Enric Meinhardt

    Accessed: 2026-04-04. Gabriele Facciolo, Carlo De Franchis, and Enric Meinhardt. Mgm: A significantly more global matching for stereovision. InBMVC 2015,

    work page 2026

  7. [7]

    An overview of the integrated software for imaging spectrometers (isis)

    Lisa Gaddis, J Anderson, K Becker, T Becker, D Cook, K Edwards, E Eliason, T Hare, H Kieffer, EM Lee, et al. An overview of the integrated software for imaging spectrometers (isis). In28th Annual Lunar and Planetary Science Conference, March 17-21, 1997, Houston, TX, p. 387., volume 28, page 387,

    work page 1997

  8. [8]

    JR Laura, Jesse Mapel, and T Hare

    doi:10.1029/2007JE003000. JR Laura, Jesse Mapel, and T Hare. Planetary sensor models interoperability using the community sensor model specification.Earth and Space Science, 7(6):e2019EA000713,

    work page doi:10.1029/2007je003000

  9. [9]

    Pengying Liu, Xun Geng, Tao Li, Jiujiang Zhang, Yuying Wang, Zhen Peng, Yinhui Wang, Xin Ma, and Qiudong Wang

    doi:10.1029/2019EA000713. Pengying Liu, Xun Geng, Tao Li, Jiujiang Zhang, Yuying Wang, Zhen Peng, Yinhui Wang, Xin Ma, and Qiudong Wang. The generation of high-resolution mapping products for the lunar south pole using photogrammetry and photoclinometry.Remote Sensing, 16(12):2097,

    work page doi:10.1029/2019ea000713 2097

  10. [10]

    Zachary M Moratto, Michael J Broxton, Ross A Beyer, Mike Lundy, and Kyle Husmann

    doi:10.3390/rs16122097. Zachary M Moratto, Michael J Broxton, Ross A Beyer, Mike Lundy, and Kyle Husmann. Ames stereo pipeline, nasa’s open source automated stereogrammetry software. In41st Annual Lunar and Planetary Science Conference, number 1533, page 2364,

    work page doi:10.3390/rs16122097

  11. [11]

    doi:10.1007/s11214-010-9634-2. D. E. Shean, O. Alexandrov, Z. M. Moratto, et al. An automated, open-source pipeline for mass production of digital elevation models (DEMs) from very-high-resolution commercial stereo satellite imagery.ISPRS Journal of Photogrammetry and Remote Sensing, 116:101–117,

    work page doi:10.1007/s11214-010-9634-2

  12. [12]

    doi:10.1016/j.isprsjprs.2016.03.012. D. E. Smith, M. T. Zuber, G. A. Neumann, et al. Initial observations from the lunar orbiter laser altimeter (LOLA). Geophysical Research Letters, 37(18),

    work page doi:10.1016/j.isprsjprs.2016.03.012 2016

  13. [13]

    doi:10.1029/2010GL043751. E. K. Stathopoulou et al. Open-source image-based 3D reconstruction pipelines: review, comparison and evaluation. In ISPRS Ann. Photogramm. Remote Sens. Spat. Inf. Sci., volume IV-2/W6, pages 331–338,

    work page doi:10.1029/2010gl043751

  14. [14]

    Bill Triggs, Philip F McLauchlan, Richard I Hartley, and Andrew W Fitzgibbon

    doi:10.1109/TGRS.2002.802878. Bill Triggs, Philip F McLauchlan, Richard I Hartley, and Andrew W Fitzgibbon. Bundle adjustment—a modern synthesis. InInternational workshop on vision algorithms, pages 298–372. Springer, 1999

    work page doi:10.1109/tgrs.2002.802878 2002