VWB is a modular, extensible computer vision framework that supports tasks including automated science and engineering analysis, large satellite image processing, and 2D/3D environment reconstruction. The framework provides a rapid C++ development environment as well as a flexible, multi-platform system to deploy computer vision applications. The module interface allows new capabilities to be rapidly integrated, and a dataflow architecture allows image processing pipelines to be quickly developed and reconfigured.
The VWB Core provides basic primitives for processing and manipulating images. The Core is built around an abstract concept of an image that decouples image processing from the underlying representation of the image itself (memory, file, procedure). VWB also includes the following modules: VWB Camera (camera models), VWB Cartography (tools for manipulating geospatially referenced images), VWB HDR (creating and compressing high dynamic range images), VWB Interest Point (tracking and matching interest points), VWB Math (geometric, numeric, and other math functions), VWB Mosaic (compositing, blending, and manipulating 2D image mosaics), VWB Plate (terabyte scale image I/O), and VWB Stereo (stereo image correlation including reference, block-based image, and multi-resolution/ pyramid-based correlators).
NASA Vision Workbench can be downloaded from https://github.com/visionworkbench/visionworkbench .