CaSSIS-based stereo products for Mars after three years in orbit

Highlights

• Pipeline for the generation of Digital Terrain Models and Anaglyphs.

• Quality assessment.

• Image content.

• Morphological characteristics.

• Comparisons with reference DTMs.

Abstract

The Colour and Stereo Surface Imaging System (CaSSIS) on board the ExoMars Trace Gas Orbiter (TGO), operating in push-frame mode, provides multiband images at four different wavelengths thanks to a Filter Strip Assembly with a panchromatic filter and three broadband filters within the visible and near infrared range. The camera acquires stereo pairs fundamental for the initialization of the photogrammetric process to perform three-dimensional reconstruction of the Martian surface at the best resolution of 4.6 m per pixel for regions up to ∼400 km² in one imaging sequence.

The 3D points derived from the stereo processing are used to generate Digital Terrain Models (DTMs) with height accuracy on the order of one image pixel on ground allowing high-resolution morphometric studies and in general improving the understanding of the geology and geomorphology of the surface of Mars.

This work provides a review of the CaSSIS stereo products supported by a description of the applied methods and examines some specific approaches directed to science analysis. Furthermore, our development of methods is herein focused on the proof of concept and the performance of our dedicated pipeline. The DTM generation procedure has been implemented in a stereo photogrammetric pipeline by the team of the National Institute for Astrophysics-Astronomical Observatory of Padova (INAF-OAPd). The workflow is based on area-based image matching integrated in a multi-resolution approach where the quality of the image matching largely determines the quality of the output DTM. For this reason, the influence of the parameters involved in the matching process (i.e. number of tie-points, template sizes and shape models in matching) has been studied.

CaSSIS stereo products have been generated for approximately 0.1% of the surface of Mars and 16.3% of the total stereo images acquired so far. In this work, some scientifically interesting targets have been considered in the investigation to provide an overview of the quality of the stereo results.

The experimental studies related to stereo analysis frequently led to comparison tests since they represent the best approach for contributing to the methodological consolidation of the photogrammetric data processing. The quality assessment based on comparison with reference terrain data is very promising also in considering areas with different surface type and morphologies.

Introduction

The Colour and Stereo Surface Imaging System (CaSSIS) (Thomas et al., 2017) with its stereo capabilities was conceived for enhancing our knowledge of the surface of Mars by extending and complementing the observations of previous instruments such as the High Resolution Imaging Science Experiment (HiRISE) (McEwen et al., 2007) which is currently operating on board NASA's Mars Reconnaissance Orbiter (MRO). Even before, other systems have imaged the surface of Mars in stereo mode. The High Resolution Stereo Camera (HRSC) (Neukum and Jaumann, 2004) (Jaumann et al., 2007); of the Mars Express mission was one of the first stereo camera designed to derive digital terrain models (DTMs) and the corresponding orthoimages on its standard operation mode with resolution better than 20 m/pixel. Furthermore, the Context Camera (CTX (Malin et al., 2007)), for other MRO observations, is able to acquire stereoscopic image pairs with imaging scale similar to CaSSIS (∼6 m/pixel) operating in push broom mode. Both HiRISE and CTX acquire stereo pairs thanks to the ability of MRO to point off-nadir such that the camera can be pointed at specific targets.

CaSSIS adopts a single focal plane with four fixed colour filters and is able to acquire along-track stereo images thanks to a rotation mechanism. In the push-frame approach, the 2D image (hereafter called “framelet”) is acquired, then buffered, read while the spacecraft moves and transferred to the proximity electronics before the next framelet is acquired, considering a sufficient surface overlap with the previous one. In the moving of the spacecraft along the orbit, CaSSIS acquires push-frame images to build up a full image swath (first of the pair) before rotating by 180°, thanks to a rotational stage, to acquire the second image of the pair. The telescope, furthermore, is mounted tilted by 10° with respect to the mechanical axis of symmetry of the bearing (nadir direction) guaranteeing a stereo convergence angle of 22.4° for a circular orbit at a height of around 400 km above Mars surface.

CaSSIS is the most recent example of a stereo camera employed in the photogrammetric mapping for planetary exploration. The processing techniques described here, applied to the imaging data, allow us to derive high-level stereo products such as the high-resolution DTMs, the corresponding orthoimages and anaglyphs. The expectations on the CaSSIS data are to provide stereo coverage over the surface equivalent to ∼1–2% of the planet for every Martian year in orbit, complementing the other stereo data in terms of high resolution, colour and 3D mapping. The availability of three-dimensional data of a planetary surface enables the quantitative morphological analysis of any particular features, such as impact crater interiors and their ejecta blankets, the volcanic domes, mountain chains or hills, mounds, drainage divides, as well as incised or inverted riverbeds and deltas.

More specifically, CaSSIS DTMs and orthoimages have been already used to investigate the stratigraphy of the South Polar Layered Deposits (Becerra et al., 2019). In addition, they have been used to perform the topographic correction on CaSSIS orthoimages (i.e., modeling and removal of brightness differences induced by topography), allowing to obtain corrected images that will be key for future photometric analyses of Maritan surface features (Munaretto et al., 2021).

The ability of CaSSIS to acquire along-track stereo, near-simultaneous stereo pairs avoids illumination differences between images and offers benefits for the DTMs and anaglyphs production.

This is especially beneficial since the orbit of TGO precesses through all local times of the day (Beta angle), and revisit times are much farther apart (depending on latitude) than those of MRO, which is in a Sun-synchronous orbit, but acquires stereo pairs on different orbits. In this paper, a description of these CaSSIS stereo products is proposed with particular attention to the effect of input parameters on the quality and precision of the DTMs. In the photogrammetric processing section, the general procedure is described and a series of different processing parameters is applied to test several strategies driven by the scientific investigation scope. The variability of the input data (image content and textures related to the illumination conditions and morphologic characteristics) suggests the application of flexible processing parameters for the derivation of highly accurate DTMs. The quality of the products has been assessed in terms of internal accuracy in terms of completeness of the surface, details of the reconstructed surface shape and morphologic consistency. The results are compared with DTMs provided by other Mars imagers (HiRISE and CTX). In order to extract robust statistics and to be able to derive considerations from the comparison a co-registration process between the datasets has been fundamental.

The paper is organized as follows. Section 2 describes the photogrammetric processing implemented in the OAPD-INAF pipeline (3DPD software). In Section 3, the CaSSIS stereogrammetric products are described and in Section 4 the anaglyph processing is presented. Section 5 elaborates the methodology and describes the experimental dataset. In Section 6, the performance analysis of the photogrammetric products is reported in details, and is also supported also by qualitative analysis. A final discussion and conclusions are provided in Section 7.