This paper describes a feature-based image registration technique that is potentially well-suited for onboard deployment. The overall goal is to provide a fast, robust method for dynamically combining observations from multiple platforms into sensors webs that respond quickly to short-lived events and provide rich observations of objects that evolve in space and time. The approach, which has enjoyed considerable success in mainstream computer vision applications, uses invariant SIFT descriptors extracted at image interest points together with the RANSAC algorithm to robustly estimate transformation parameters that relate one image to another. Experimental results for two satellite image registration tasks are presented: (1) automatic registration of images from the MODIS instrument on Terra to the MODIS instrument on Aqua and (2) automatic stabilization of a multi-day sequence of GOES-West images collected during the October 2007 Southern California wildfires.
Onboard Image Registration from Invariant Features
i-SAIRAS: International Symposium on Artificial Intelligence, Robotics and Automation in Space ; 2008 ; Los Angeles, CA, United States
26.02.2008
Aufsatz (Konferenz)
Keine Angabe
Englisch
ENCRYPTION KEY REGISTRATION SYSTEM, ENCRYPTION KEY REGISTRATION METHOD, AND ONBOARD DEVICE
Europäisches Patentamt | 2023
|NTRS | 1980
|Distinctive Image Features from Scale-Invariant Keypoints
British Library Online Contents | 2004
|NTRS | 1979
|