An Embedded System for Aerial Image Processing from Unmanned Aerial Vehicles

• Jonas Fernandes da Silva UFPB
• Alisson V. Brito UFPB
• Jose Antonio Gomes De Lima UFPB
• Helder Nogueira De Moura REINTEGRIS

Resumo

This paper proposes solution capable to process aerial images from UAVs to identify failures in plantations and makes a comparison of the system running on light sized computers and low power computing platforms. An algorithm was developed based on watersheds using OpenCV library. The solution was embedded on X86 architecture (AlteraDE2i-150) and Intel Edison boards as well as on ARM architecture (Raspberry Pi 2). The results show that the proposed system is a cost-effective solution for the problem of fault identification in plantations, and can be embedded in UAVs for processing images in real time.

Palavras-chave: Power demand, Embedded systems, Computer architecture, Real-time systems, Monitoring, Image segmentation, UAV, Watersheds