Fig. 2: Integration Principle.
From: An autonomous drone swarm for detecting and tracking anomalies among dense vegetation
a Integration principle illustrated with three drones (D1, D2, D3) for occlusion suppression in a defined synthetic focal plane (SA imaging) and for the suppression of registration artifacts due to unknown heading parameters (registered vs. unregistered). b Simulation example for 5 drones (D1-D5) and an occluded person lying in a forest (300 trees/ha). Single RGB images (left column). Corresponding single anomaly images (center column). Occlusion-suppressed anomaly integrals and blob detection results for the cases of precise registration and for the case of imprecise registration (due to heading errors) after integrating 10 heading variations in 1° steps (right column). Details of the simulator are provided27. c Real recordings by three drones of three unoccluded circular targets on the ground. Top, middle, bottom: RGB integrals after precise initial registration and after misregistration due to compass drift. Anomaly integral after compass drift. Anomaly integral including heading integration after compass drift and final blob detection.
