Towards a fast image alignment technique for multi-aperture UAV and CubeSat payloads

This paper compares various image alignment techniques implemented on low-powered embedded devices for use in multi-aperture UAV and CubeSat payloads. A novel method was introduced and executed alongside two established methods on an embedded processor: utilising images taken from a dual-aperture UAV payload for the performance evaluation. The techniques were evaluated based on processing time, wherein significant differences were observed, outside of the universally comparable structural similarity index (SSIM). The outcomes of this research indicate that among the methods compared, the proposed method requires the least amount of processing time without sacrificing alignment quality, making it suitable for use in, applicable, low-power applications.