Background
Among the first steps of any planetary surface investigation is to construct a mosaic, which facilitates innumerable geologic and geophysical studies. This holds regardless of whether mosaics already exist: Ever-higher-resolution mosaics enable new studies and almost always quickly supersede the previous generation. We worked to develop workflow to create global 6 meters/pixel Mars Reconnaissance Orbiter (MRO) ConTeXt (CTX) camera mosaics of the planet Mars.
A mosaic currently does not exist because (a) enough images only recently became available, (b) the images are difficult to process, and (c) the sheer number of images to mosaic is daunting. The processed images alone total 13 terabytes of data, making this both an optimization and data processing problem. We were awarded two Quick-Look proposals to develop, refine, and work through remaining issues in developing this workflow in order to make the process feasible and potentially fundable by federal agencies.
An important component of this is that the mosaicking process is not as simple as just putting multiple images together; rather, because the exact spacecraft pointing is uncertain and has built-in errors, we must first develop a process to determine where the spacecraft and camera was pointed for each observation. We can then use that solution to re-project images to where they should be on the planetary surface and mosaic them together.
Approach
We began with a basic, iterative process of creating semi-automated control networks. A control network takes multiple images, identifies common features, and determines how the a priori spacecraft and camera pointing information must be altered in order to have those features project to the same location. Once that is solved, the images can be re-projected with our new solution and mosaicked. The mosaicking itself has issues due to properly tone-matching images to each other, and ensuring the best images are placed on top due to varying planetary conditions or spacecraft faults. Our initial approach was to have this part of the process be fully manual.
Accomplishments
We succeeded in developing a solid workflow that works for Mars CTX images. We are cautious here as to how well this workflow might be generalized to other data and other bodies because the specific camera and imaging campaign for MRO CTX are ideal for creating even, high quality mosaics.
We modified our original approach to remove much of the iterative component, which greatly reduced the time needed by the computer (from approximately one week per region to half a day). Additionally, we implemented multiple and different methods for the computer to use to try to match features between images to create a good, accurate control network. This significantly reduced the amount of manual input required in creating the control networks, in many cases from over 20 hours per region to just 1 hour. Additionally, we developed a semi-automated method of creating the image order to mosaic the data and found tools to tone-match images in an automated way, further reducing the manual labor required.
Overall, we went from a process that mostly worked to one that fully worked, and one that would have required about 7000 manual hours to one that we proposed to NASA’s The Planetary Data Archiving, Restoration, and Tools (PDART) program that would take about 2500 manual hours.
This project is associated with High-Resolution Controlled ConTeXt Camera Image Mosaics of Mars, 15-R8809.