Algorithms for Stereo Image Creation from a Sequence of Two Dimensional Images

Stellar Science has developed innovative techniques capable of autonomously converting standard 2D image sequences into 3D stereo movies, and has successfully tested these techniques on a wide variety of space surveillance imagery. Our algorithm first solves for the satellite’s shape and motion using a feature based technique, and then uses this information to select and combine suitable stereo image pairs, and finally when desired, generates realistic looking synthetic images to increase the temporal fidelity of the image sequences. In phase II, we will build upon and expand this initial capability by investigating specific techniques that promise to improve the quality, quantity and usefulness of our output stereo images. Our phase II plan includes creating synthetic images that do not lie in the imaging sensor’s viewing baseline, implementing a detailed testing mechanism that can characterize these improvements, supporting new space and non-space related imagery types such as radar, and demonstrating software tools that help users scientifically exploit and measure stereo imagery. We will combine our vast experience in space surveillance imagery exploitation and stereo image creation techniques with our proven track record of implementing robust scientific algorithms to deliver a working and highly maintainable software tool.