Interactive Atmospheric Turbulence Mitigation
Dmitri Kamenetsky, Michael Zucchi, Geoff Nichols, David Booth, Andrew J. Lambert · 2016
Surveillance imagery acquired at long distances is frequently degraded by atmospheric turbulence, causing it to be blurry and distorted. We introduce an interactive method for atmospheric turbulence mitigation of video sequences. Our method allows the user to input crucial information about the imaging conditions, hence bypassing many of the complexities involved in other approaches. Experiments on real-world data show that our method strikes a good balance between speed and accuracy, and compares favourably to state-of-the-art methods, such as CLEAR. We also introduce a new image quality assessment technique suitable for turbulence affected imagery. Our technique is the first to use partial-reference assessment, where ground truth is available for only one of the objects in the scene.