REAL-TIME IMAGE PROCESSING IMPLEMENTATION FOR ON-BOARD OBJECT DETECTION AND TRACKING

Resident Space Objects (RSOs) monitoring has become one of the main goals for many projects in the framework of Space Situational Awareness (SSA) and Space Surveillance and Tracking (SST). The Star sensor image on-board Processing for orbiting Objects deTection (SPOT) fits in this field providing an innovative space-based autonomous and versatile solution for RSOs’ optical detection via star sensors and for different Earth orbits scenarios. This system is being designed and it is part of the In-Orbit Validation (IOV) activity in the near future. The SPOT algorithms involve many complex arithmetic operations, which are relatively expensive in terms of computational latency, limiting their applicability to real-time image processing applications. The purpose of this paper is to show the Software / Hardware co-design methodology and the implementation of SPOT modules on a Zynq-7000 SoC (System-on-Chip) that integrates a dual-core ARM Cortex-A9 processor associated with an FPGA (Field Programmable Gate Array). Thus, the co-design approach makes them work together to share the processing load and enables high-speed image flow processing. Several methods have been applied to find a trade-off between resources utilization and energy consumption with realtime requirements. A Hardware-in-the-Loop (HIL) setup was developed as well, to evaluate the performance and robustness of the SPOT algorithms and simulate critical scenarios. Results conclude that the hardware SPOT implementation accelerates the execution speed approximately by 70 times with respect to the Intel CPU implemented software solution.