Foreground segmentation in atmospheric turbulence degraded video sequences to aid in background stabilization

Citation:

P.E. Robinson and A.L. Nel, “Foreground segmentation in atmospheric turbulence degraded video sequences to aid in background stabilization”.

Abstract:

Stabilization of the geometric distortions present in atmospheric turbulence degraded (ATD) video sequences containing objects undergoing real motion is a challenging task. This is due to the difficulty of discriminating what visible motion is real motion and what is caused by ATD warping. Due to this most stabilization techniques applied to ATD sequences distort real motion in the sequence. In this study we propose a new method to classify foreground regions in ATD video sequences. This classification is used to stabilize the background of the scene while preserving objects undergoing real motion by compositing them back into the sequence. A hand annotated dataset of three ATD sequences is produced with which the performance of this approach can be quantitatively measured and compared against the current state-of-the-art.

Dataset:

The following dataset is provided as part of this paper and includes 100 hand annotated frames from 3 atmospheric turbulence degraded video sequences. More details on the segmentation encoding and sequences are available in the paper. The three ATD sequences were graciously provided by the CSIR of South Africa’s Optronic Sensor Systems group with the financial support of Armscor.

If you make use of this dataset please cite the above article as the source.

ATD_Dataset.zip

Supplementary Material:

The following material is high-resolution images and video that supports this article and could not be adequately reproduced in the publication format.

Supplementary Material (zip)