Structure from motion using optical flow probability distributions

Paul Clark Merrell; Dah-Jye Lee

doi:10.1117/12.604535

Translator Disclaimer

28 March 2005 Structure from motion using optical flow probability distributions

Paul Clark Merrell, Dah-Jye Lee

Author Affiliations +

Proceedings Volume 5803, Intelligent Computing: Theory and Applications III; (2005) https://doi.org/10.1117/12.604535
Event: Defense and Security, 2005, Orlando, Florida, United States

Abstract

Structure from motion is a technique that attempts to reconstruct the 3D structure of a scene from a sequence of images taken from a camera moving within the scene. Structure from motion can be used on an Unmanned Aerial Vehicle or Unmanned Ground Vehicle for obstacle detection as well as for path-planning and navigation. The 3D structure of the scene is estimated using the optical flow values found at a set of feature points on the image. Typically, this is done under the assumption that all of the optical flow values have the same level of accuracy and that the accuracy of each optical flow value in the horizontal and vertical directions is the same. These assumptions are not entirely correct. The accuracy of different optical flow values are not the same and the accuracy can be measured and quantified using an optical flow probability distribution. We will present a novel structure from motion algorithm that is more accurate and more robust than other methods because it uses optical flow probability distributions. We will present one method that is designed to work on only two frames and another method designed to work on any number of frames.

Citation Download Citation

Paul Clark Merrell and Dah-Jye Lee "Structure from motion using optical flow probability distributions", Proc. SPIE 5803, Intelligent Computing: Theory and Applications III, (28 March 2005); https://doi.org/10.1117/12.604535

ACCESS THE FULL ARTICLE