Detection and classification of moving objects from UAVs with optical sensors

Abstract

Small and medium sized UAVs like German LUNA have long endurance and define in combination with sophisticated image exploitation algorithms a very cost efficient platform for surveillance. At Fraunhofer IOSB, we have developed the video exploitation system ABUL with the target to meet the demands of small and medium sized UAVs. Several image exploitation algorithms such as multi-resolution, super-resolution, image stabilization, geocoded mosaiking and stereo-images/3D-models have been implemented and are used with several UAV-systems. Among these algorithms is the moving target detection with compensation of sensor motion. Moving objects are of major interest during surveillance missions, but due to movement of the sensor on the UAV and small object size in the images, it is a challenging task to develop reliable detection algorithms under the constraint of real-time demands on limited hardware resources. Based on compensation of sensor motion by fast and robust estimation of geometric transformations between images, independent motion is detected relatively to the static background. From independent motion cues, regions of interest (bounding-boxes) are generated and used as initial object hypotheses. A novel classification module is introduced to perform an appearance-based analysis of the hypotheses. Various texture features are extracted and evaluated automatically for achieving a good feature selection to successfully classify vehicles and people

Multi-modal target detection for autonomous wide area search and surveillance

“…This is used as an empirical test case to explore the optimization of an observation model for the environment when distributed over multiple sensors, platforms and resulting target detection signatures. With an ever increasing use of remote deployed autonomous systems the problem of reviewing, processing and effectively reporting the sensor information they gather is one of growing significance with notable parallels in ground-based sensor networks [9,10] Prior work in the field is generally limited to the case of automated ground surveillance (the camera network case [9,[11][12][13][14]), isolated to the UAV based detection [5,7,[15][16][17][18] with only very limited wide-scale consideration of multi-platform dual aerial and ground platforms within the same sensing scenario [19,20]. Notably prior work does not address autonomy within this context [19,20] and work considering multi-modal detection is in its infancy [7,17,18].…”

REDBEE: A visual-inertial drone system for real-time moving object detection

“…Chong Huang, Peng Chen, Xin Yang, and Kwang-Ting (Tim) Cheng, Fellow, IEEE surveillance system on an embedded system [20,21,22]. These approaches are based on the assumption that the background can be approximated by a plane.…”

“…If the drone performs the detection during flight, the camera motion compensation is necessary for the background modeling. Michael et al [20] developed the video exploitation drone system ABUL for detection and classification of moving objects. Their system relies on fast and reliable video data transmission between the drone and the ground station.…”

“…Most of the real-time applications [20] on camera drones involve sending a video data stream back to a ground station with a high-end computing server. However, a fast and reliable data connection between the drone and the ground station is essential.…”

Recognizing Humans in Motion: Trajectory-based Aerial Video Analysis

“…Leitloff et al proposed a method to detect cars by adapting boosting in combination with Haar-like features [8], and Schmidt et al proposed a slightly similar method with [8] to detect people based on Haarlike features [13]. Teutsch et al extracted appearance features, such as moments and local binary pattern (LBP), with a 9-NN classifier [14]. In these methods the image resolution of object area is large enough to obtain appearance information, so objects were detected successfully.…”