Evaluating the Computational Capabilities of Embedded Multicore and GPU Platforms for On-Board Image Processing

On-board Image Processing is one of the most computationally intensive tasks performed in space payload processing. With the constant increase in the resolution of optical sensors, new powerful architectures are needed for their processing. In this work, we evaluate the capabilities of several state-of-the-art embedded COTS and radiation tolerant multicore and GPU featuring platforms, with an open source on-board image processing application we have developed and parallelised. Our results show that both embedded multicores and especially GPUs are very effective for such tasks. In fact, GPUs become even more efficient when sensor sizes are increasing, making them ideal candidates for future space missions.