Spatio-temporal Super-Resolution of Thermal Image Sequences through Kalman Filtering and Smoothing

This paper addresses the challenge of spatio-temporal enhancement of thermal imaging in additive manufacturing (AM), where conventional thermal imaging systems struggle with limited spatial and temporal resolution. A novel approach is proposed to extend the Kalman filtering and Smoothing framework for combined spatial and temporal super-resolution, treating the problem as a missing observation estimation task. The method integrates a state-space representation of the thermal diffusion process, leveraging a Kalman filter and smoother to predict inter-frame thermal variations and refine spatial details. Simulation results demonstrate that the proposed approach effectively reconstructs high-resolution thermal images while preserving transient thermal behaviors, outperforming bi-cubic interpolation in terms of reconstruction quality.