Dynamic Distributed-Aperture Computational Imaging For Remote Sensing


The increasing use of drones across various industries is revolutionizing large-scale mapping, with their small size, lightweight design, and agility enabling the creation of drone swarms to cooperatively complete complex missions. These missions range from surveillance and reconnaissance to target search, tracking, and mapping.


This project formulates a new distributed aperture imaging concept for a coordinated fleet of mapping platforms, synthesizing a unified virtual tiled focal plane array from multiple sensor focal planes.


Trade-offs between coverage and sampling resolution in aerial imaging 

The diagram below illustrates the fundamental trade-off in aerial imaging between coverage area and ground sampling resolution (GSD). The key parameters include the flight altitude (h), sensor width (w_s), focal length (f_s), and sensor pitch (p_s). As the altitude increases, the field of view (FOV) and ground coverage width (w) expand, offering broader scene coverage. However, this comes at the cost of reduced ground sampling resolution (GSD), which defines the pixel size on the ground. The formula for GSD reflects the relationship between sensor characteristics and flight parameters, emphasizing how adjustments in altitude or sensor properties affect imaging performance.

aaa

Integrating multiple focal planes to enhance camera coverage while preserving a desired GSD

Merging the focal planes of multiple cameras is an effective method with the following benefits:

Related works

Limitations of Current Aerial Multi-Camera Systems


Misalignment in a rigid multi-camera setup, influenced by scene variations, altered flight parameters, and 3D parallax effects.

Disruption in a 2-camera setup: Effects of vibration and mechanical disturbances

Simulation Using Real Images and Sparse Point Cloud


Real Image Using a 4-Camera Array Hardware

Simulation Using BA4S (SfM) and Unreal Engine (UE)

Dynamic Distributed-Aperture Computational Imaging Using A Fleet of UAVs

Drone swarm configurations for imaging demonstrating reconfigurability and flexibility in our approach

Our innovative approach harnesses the reconfigurability of drone clusters to establish a versatile distributed aperture computational imaging system, creating an expansive virtual focal plane in the sky.

Virtual Focal Plane Array and Reprojection in Our Method

Reference

H. Aliakbarpour, J. Collins, E. Blasch, V. Sagan, R. Massaro, G. Seetharaman, and K. Palaniappan, "Flexible Multi-Camera Virtual Focal Plane: A Light-Field Dynamic Homography Approach", In Scanning Technologies for Autonomous Systems. Springer Nature, 2024. pdf download bibtex