Abstract
Small unmanned aircraft vehicles (SUAVs) or drones are very useful for visual detection and tracking due to their efficiency in capturing scenes. This paper addresses the detection and tracking of moving pedestrians with an SUAV. The detection step consists of frame subtraction, followed by thresholding, morphological filter, and false alarm reduction, taking into consideration the true size of targets. The center of the detected area is input to the next tracking stage. Interacting multiple model (IMM) filtering estimates the state of vectors and covariance matrices, using multiple modes of Kalman filtering. In the experiments, a dozen people and one car are captured by a stationary drone above the road. The Kalman filter and the IMM filter with two or three modes are compared in the accuracy of the state estimation. The root-mean squared errors (RMSE) of position and velocity are obtained for each target and show the good accuracy in detecting and tracking the target position—the average detection rate is 96.5%. When the two-mode IMM filter is used, the minimum average position and velocity RMSE obtained are around 0.8 m and 0.59 m/s, respectively.
Highlights
The Kalman filter and the Interacting multiple model (IMM) filter with two or three modes are compared in the accuracy of the state estimation
The use of small/miniature unmanned aerial vehicles (SUAV) or drones has increased for a variety of applications
Several moving people and cars were captured by an SUAV
Summary
The use of small/miniature unmanned aerial vehicles (SUAV) or drones has increased for a variety of applications. The centroids of final region of interest (ROI) windows are considered x and y positions, which are fed to the tracking stage as measurements This detection approach does not require intense training process and heavy computational burdens. The major contributions of this paper lie in the following: (1) we integrate visual detection based on image processing and a target tracking derived from statistical estimation. It is noted that we instantly get dynamic state estimates, such as position, velocity, and acceleration, in the proposed method; (2) No massive training data is required for target detection and tracking. This method can speed up the process, with less computational resources.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
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 callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
