Compensation of temperature effects on imaging quality of thermal imaging objectives using deep learning techniques

Abstract

Thermal imaging objectives are made from infrared materials with large thermal expansion coefficients, such as Ge, Si, and ZnSe. When the temperature changes, it leads to variations in the refractive index, curvature radius, and thickness of the lens, causing defocus shifts that degrade the image quality of the thermal imaging system. In this paper, we propose a novel method to compensate for the effects of temperature variations on the quality of thermal imaging objectives by using deep learning techniques. The temperature variations are measured using a thermal sensor. Subsequently, a U-Net network is employed to mitigate the impact of temperature on the imaging quality of the thermal imaging objectives without requiring any optical displacement or replacement of the lens. Simulation results show that the proposed method performs the effectively compensation for the influence of temperature changes on thermal imaging objective over a wide temperature range from -5 °C to 50 °C.