High-Performance Image Processing and Cartography using the NASA Vision Workbench

Presentation | Presented

  • Matthew Hancher, NASA Ames Research Center

The NASA Vision Workbench is a newly-released open-source C++ image processing library that builds on over a decade of image processing expertise at the NASA Ames Research Center and is geared especially toward geospatial image processing and robot vision. It greatly simplifies the process of developing custom high-performance image processing pipelines, and includes powerful support for cartography, automated image alignment, 3D stereo reconstruction, and processing of huge images, among other things. It also includes a set of Python bindings to simplify integration with Python-based toolchains.

We will present the Vision Workbench in the context of NASA applications, such as generating high-resolution Lunar and Martian maps or terrain models and reprocessing Earth-observing satellite imagery to produce educational and outreach content. We will also discuss how the Vision Workbench is integrated at Ames into an broader image processing and visualization toolkit based on international and de facto standards such as WMS and KML, in conjunction with other open-source image processing and mapping software. Finally, we will discuss the plans for the future of the Vision Workbench as well as challenges to open-source development in a mixed government-employee and contractor setting.