Abstract
Image stacking is a well-known method that is used to improve the quality of images in video data. A set of consecutive images is aligned by applying image registration and warping. In the resulting image stack, each pixel has redundant information about its intensity value. This redundant information can be used to suppress image noise, resharpen blurry images, or even enhance the spatial image resolution as done in super-resolution. Small moving objects in the videos usually get blurred or distorted by image stacking and thus need to be handled explicitly. We use image stacking in an innovative way: image registration is applied to small moving objects only, and image warping blurs the stationary background that surrounds the moving objects. Our video data are coming from a small fixed-wing unmanned aerial vehicle (UAV) that acquires top-view gray-value images of urban scenes. Moving objects are mainly cars but also other vehicles such as motorcycles. The resulting images, after applying our proposed image stacking approach, are used to improve baseline algorithms for vehicle detection and segmentation. We improve precision and recall by up to 0.011, which corresponds to a reduction of the number of false positive and false negative detections by more than 3 per second. Furthermore, we show how our proposed image stacking approach can be implemented efficiently.
Highlights
Aerial videos acquired by airplanes or fixed-wing unmanned aerial vehicles (UAVs) are a good source of data for ground surveillance
In the second module that is depicted in red color, these motion clusters are used as regions of interest (ROIs) to detect and segment individual moving objects
While the first three sequences SEQ 1, SEQ 2, and SEQ 3 are coming from your own top-view aerial video data, the fourth is the EgTest[01] sequence taken from the publicly available video verification of identity (VIVID) dataset.[56]
Summary
Aerial videos acquired by airplanes or fixed-wing unmanned aerial vehicles (UAVs) are a good source of data for ground surveillance. Camera and objects are moving, the observed vehicle’s appearance is nearly constant over time due to the top-view camera angle and due to the large distance between camera and scene In this way, stationary image content is removed that can modify the observed object’s appearance and, can disturb the detection and segmentation process, such as (1) partial occlusions by trees, power supply lines, or buildings, (2) stationary objects close to the observed object, such as parked vehicles or buildings, and (3) street textures, such as cobblestones or road markings. Object detection and segmentation algorithms are implemented, improved by our image stacking approach, and evaluated using FMV sequences that contain occlusions, parked vehicles, and street textures.
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
