Compact Physics-Informed Photo–Thermal Crosstalk Model for Multichip LED Display Modules

This article presents a physics-informed framework to model and efficiently predict 2-D photo–thermal crosstalk in multichip light-emitting diode (LED) display modules. A closed-form opto-thermal decomposition is derived that separates direct photoluminescent gain from indirect photo$\boldsymbol {\to }$thermal attenuation and explicitly represents thermal coupling, photoabsorption, and the temperature sensitivity of emission. The model is calibrated with microscopic hyperspectral imaging ($\boldsymbol {\mu }$-HSI) and compressed into a hardware-friendly$7~\boldsymbol {\times }7$order-space grayscale–spatial polynomial surrogate fit by constrained least squares with spatial regularization. Experiments reproduce the nonlinear crossover from optical-dominated to thermal-dominated crosstalk across the full 0–255 gray range. The block surrogate attains average relative errors below 1% (localized maxima up to 11%) and enables accurate pre-compensation and actionable diagnostics for thermal and optical design.