NASA Logo, National Aeronautics and Space Administration

+NASA Home

+Ames Home

Intelligent Robotics Group Releases High-Resolution Lunar Maps Generated by New Computer Vision Algorithms
+ Home + Organization + News + Research Areas + Publications + Software Mgmt. Office + Internal Systems Help
Intelligent Systems Division Banner

Intelligent Robotics Group Releases High-Resolution Lunar Maps Generated by New Computer Vision Algorithms

The Intelligent Robotics Group has released new high-resolution maps of the lunar surface that were created from over 4,000 images taken by the Apollo Metric (Mapping) Camera, which flew aboard Apollo’s 15, 16, and 17. The “Apollo Zone” Digital Image Mosaic (DIM) and Digital Elevation Model (DEM) cover approximately 18% of the Lunar surface at a resolution of 1024 pixels per degree (approximately 30 m/pixel). The maps are the result of three years worth of work by the Intelligent Robotics Group (IRG) to align and process thousands of images.

The “Apollo Zone”maps cover the following sites of interest: Apollo 15, Apollo 16, Alphonsus Crater, Rima Prinz, Aristarchus Plateau-2, Ina D Caldera, Sulpicius Gallus, Mare Crisium, Mare Smythii, King Crater, Tsiolkovskiy Crater, Aitken Crater, and half of the Van de Graaf Crater. The terrain model has an average vertical accuracy of 40 m/pixel and a standard deviation of 37 m (compared to LOLA laser altimetry tracks). Over 46% of the covered surface has vertical errors lower than 25 m.

BACKGROUND: The “Apollo Zone” maps (image, elevation, hillside, colorshade, confidence, and precision) were automatically generated using new computer vision algorithms developed by the Intelligent Robotics Group (IRG):

  • Robust statistical sub-pixel stereo correspondence
  • Robust bundle adjustment and radiometric corrections for large-scale image mosaics
  • Orbital camera position/orientation estimation using interest point extraction
  • Photometric correction of exposure time, shadow removal, and generation of seamless large-scale image mosaics
  • Photometric method for reconstructing lunar albedo
  • Photoclinometric terrain reconstruction method that improves lunar DTM precision
  • Statistical method for multiple stereo digital terrain-model mosaicing
  • Multi-view 3D terrain reconstruction
  • DTM/LOLA alignment and lidar/image matching

These algorithms have been released as a NASA Open Source release (Ames Stereo Pipeline, Neo-Geography Toolkit, and NASA Vision Workbench). Map processing was performed using the NASA Pleiades supercomputer. In addition to the Apollo Metric Camera images, the fully automatic map processing pipeline has also been used with data from the Lunar Reconnaissance Orbiter Camera (LROC) and by several planetary science groups.

NASA PROGRAM FUNDING: Lunar Mapping and Modeling Project (LMMP)

COLLABORATORS: NASA Marshall Space Flight Center (MSFC), NASA Goddard Space Flight Center (GSFC), the Jet Propulsion Laboratory (JPL), and the United States Geological Survey (USGS)

Contact: Terry Fong

First Gov logo
NASA Logo -