Nicholas Owen Ralph, Western Sydney University; Darren Maybour, Western Sydney University; Alex Marcireau, Western Sydney University; Imogen Jones, Western Sydney University; Ain De Horta, Western Sydney University; Gregory Cohen, Western Sydney University
Keywords: Neuromorphic, Event-Based Camera, Space Situational Awareness, Astronomy, Artificial Contract Generation
Abstract:
Neuromorphic Event-Based Cameras (EBC) are bio-inspired CMOS cameras that detect changes in luminance. They are characterised by an incredibly high temporal resolution (in the order of a few microseconds) and low power consumption. The EBC has been successfully demonstrated as a novel and effective approach to Space Situational/Domain Awareness (SSA/SDA) and space imaging, capable of observing and characterising satellites with a significantly higher temporal resolution and lower output data volume than conventional CCD sensors, which require longer exposure times and suffer from blur effects when observing moving targets.
Yet, the EBC suffers from the ‘perfect tracking phenomena, where faint stars and satellites may only become visible with contrast generated either by mutual motion between the source and telescope, atmospheric seeing, or improper tracking. Without significant changes in the incoming light from a target, there may be little detectable contrast for the EBC, especially for faint targets. Although unintuitive, a solution to this challenge is to re-introduce a controlled mutual motion into the imaging system to generate additional detectable contrast. Due to the high-speed performance of these sensors, EB space imaging systems can embrace this antithesis of conventional space imaging with improved target detection with less stable tracking.
In this paper, an Artificial Contrast Generator (ACG) is developed to simulate mutual motion in the EBC, similar to saccadic eye motion in biological vision systems for improved target detection. The ACG translate incoming light from the telescope across the EBC detection surface in a defined pattern to visualise an otherwise static scene. By inducing features on the image plane with a known direction and magnitude, actual emission can also be more easily differentiated from noise through the correlated motion using a correction algorithm. The ACG will be evaluated by comparing typical EBC and ACG augmented observations using real-world data of dense star fields using the Astrosite, a containerised observatory. Improvements to the position error, pixel scale, detected source count, and the faintest detectable sources will be estimated by comparing astrometrically calibrated observations using standard EBC and ACG augmented data collection.
The ACG is expected to produce significantly more position samples of a visible target by creating contrast in multiple pixels, which can be analysed statistically to estimate a sub-pixel target position. Furthermore, the ACG is expected to improve the Signal-to-Noise Ratio of an EB space imaging system, address the perfect tracking phenomena and improve the spatial resolving capabilities of an EB space imaging system.
The ACG device and correction algorithm developed in this paper will enable EB space imaging systems to better serve the goals of SSA in providing accurate and timely information on the space domain with improved sensitivity and spatial accuracy. The ACG imaging techniques and algorithms developed in this paper will provide the space-sensing field with a deeper understanding of the EB vision paradigm. The outcomes of this paper continue efforts to improve the capabilities of EBCs for use in tracking and space imaging tasks and therefore contribute to the growing efforts of the international SSA and the development of the EB technology in astronomy and space science applications.
Date of Conference: September 19-22, 2023
Track: SDA Systems & Instrumentation