Abstract
Vehicle surveillance of a wide area allows us to learn much about the daily activities and traffic information. With the rapid development of remote sensing, satellite video has become an important data source for vehicle detection, which provides a broader field of surveillance. The achieved work generally focuses on aerial video with moderately-sized objects based on feature extraction. However, the moving vehicles in satellite video imagery range from just a few pixels to dozens of pixels and exhibit low contrast with respect to the background, which makes it hard to get available appearance or shape information. In this paper, we look into the problem of moving vehicle detection in satellite imagery. To the best of our knowledge, it is the first time to deal with moving vehicle detection from satellite videos. Our approach consists of two stages: first, through foreground motion segmentation and trajectory accumulation, the scene motion heat map is dynamically built. Following this, a novel saliency based background model which intensifies moving objects is presented to segment the vehicles in the hot regions. Qualitative and quantitative experiments on sequence from a recent Skybox satellite video dataset demonstrates that our approach achieves a high detection rate and low false alarm simultaneously.
Highlights
An increasing number of commercial earth observation satellites have been launched over the past decades that generate large quantities of satellite imagery with resolution of 1 m (Ikonos) or even better (QuickBird, geoEye), and most related satellite imagery based works experiment on their imagery
On accounting for that it seems more attractive of vehicles in the local region than the global image in the satellite video sequences, local saliency is employed to intensify the objects as the first step of background model
We identify the detection results as follows: for each connected domain in the motion detection result, the one which has at least one pixel overlap with ground truth is supposed to be a true positive (TP); otherwise, it is regarded as a false positive (TP), and false negative (FP) is the moving vehicle which has not been found
Summary
An increasing number of commercial earth observation satellites have been launched over the past decades that generate large quantities of satellite imagery with resolution of 1 m (Ikonos) or even better (QuickBird, geoEye), and most related satellite imagery based works experiment on their imagery. No available appearance or shape information can be extracted to carry out a common classification algorithm Another difficulty for the vehicle detection lies in the much more complicated background of the satellite videos. Besides the common objects in aerial image, like varying magnitudes of buildings and trees, the slowly moving airplanes can be found in the satellite video sequences. Though less appearance information of vehicles can be utilized for detection, some methods are still proposed for object detection in high-resolution satellite imagery. For the satellite video sequence, the resolution is limited so that common shape or appearance based methods for object detection no longer make sense. We conduct experiments on the satellite video sequence using the proposed method, and results show a high detection rate with a low false alarm rate.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
