AIFLY

Infrared imaging plays a pivotal role in the surveillance and monitoring of small unmanned aerial vehicles (UAVs). However, the images captured by these systems often contain substantial noise, which can interfere with effective analysis. To address this challenge, we have focused our attention on long-wavelength infrared (LWIR) cameras, specifically those that operate within the 8 μm–14 μm range. These cameras, when paired with polarization filters, provide improved clarity by reducing noise in the infrared spectrum. However, the inherent noise in these images is compounded by two major sources: polarization noises and vibrational disturbances caused by the UAV's 6 degrees of freedom (DoF). While a variety of Artificial Intelligence-based algorithms have been developed to tackle polarization noise, the real challenge emerges when these solutions are applied to moving UAVs. The UAVs' vibrations, which are a result of their motion, add a layer of complexity to real-time image processing. Addressing these dual sources of noise—polarization and vibrational—has been a major focus in the development of our Artificial Intelligence-powered octocopter, setting it apart from competing systems on the market.