DESIGN AND IMPLEMENTATION OF REAL-TIME TURBULENCE MITIGATION AND CHARACTERIZATION SYSTEMS

Abstract

Atmospheric turbulence can be caused by differential heating of the air and the ground. The differential heating effect causes localized changes in the refractive index and thus wavefront, resulting in distortion and loss of resolution in imagery taken over long ranges, limiting the actionable information contained in the scene. Correcting imagery degradation rapidly enough to improve real-time long-range imaging use cases such as flight testing or target acquisition requires low latency, fast implementations, and a high degree of system and scene independence. Development of a real-time turbulence mitigation solution based on augmenting existing algorithms for performance and real-world usability, and creating real-time implementations using accelerated computing technologies are discussed. Several candidate algorithms are evaluated for real-time image enhancement and compared with a novel method for comparing image quality. The chosen algorithm implementation is also evaluated based on the benefit it provides to image exploitation tasks such as facial recognition and target acquisition. Atmospheric turbulence parameters such as aggregate strength and wind velocity can be used to inform performance predictions, as well as correct and optimize the performance of imaging systems, optical communication and measurement systems, and weapons systems in addition to their use in correcting image degradations. Real-time algorithms for measuring atmospheric turbulence strength in real-world imaging scenarios are subject to the same constraints as image enhancement algorithms. This dissertation discusses creating an algorithm for measurement of atmospheric turbulence strength that meets these constraints. Wind velocity measurement has many uses outside of atmospheric turbulence correction but can be measured by observing the temporal behavior of atmospheric turbulence. A system for real-time measurement of wind velocity from passive imagery from a single aperture imaging system based on observing arbitrary targets is presented.