Abstract
Artificial motion and warping of images taken at long range is one of the most significant and troublesome effects of atmospheric turbulence. It is important to understand and model this effect correctly in order to: 1) fully characterize turbulence between the target and the observer, 2) devise efficient post-processing strategies for artificial motion correction, and 3) exploit information about statistics of this atmospheric motion to distinguish between real and fake movement in a scene. This paper discusses two types of motion: G-tilt and Z-tilt, highlighting the differences between them. Optimal image block size for de-warping algorithms and bandwidth considerations are given special attention. Finally, strategies for turbulence characterization based on differential image motion are discussed.