Stellar Science's image processing and computer vision capabilities span the gamut from image enhancement to 3D object reconstruction and rendering.
Many of our image processing applications are designed to let users fully exploit scientific imagery, including 16-bit, floating point, and multi-spectral images, along with any associated metadata. Our cross-platform image processing library includes a pipeline that allows users or to instantly sharpen, gamma adjust, equalize, color map, interpolate, and align selected channels of an image in order to effectively visualize an object of interest.
We also offer a number of Photoshop plugins to read and write scientific image formats and have authored an open source library for parsing NITF 2.0 files.
A major focus of our computer vision research and development efforts is the computation and exploitation of 3-dimensional structure from 2-dimensional imagery. Computer vision applications that we have developed include:
- Stellar Shape Project, a shape-from-motion application that computes 3D textured models of satellites from 2D imagery taken from a variety of viewpoints.
- Stereo Image Creator (sic), an application that automatically synthesizes stereoscopic 3D images of an object from 2D imagery, including 3D images from novel viewpoints.
- Caliper, a mensuration application that allows an image analyst to accurately measure an object in 3D using 2D imagery.
- Facet to Model, an application to automatically fit polynomial and NURBS surfaces to triangulated 3D models.
We have implemented capabilites to infer materials from multispectral imagery and algorithms for rendering thermal data; we have also analyzed microscopic images of test materials for evidence of thermal damage.