Full-chain model-driven cross-level multi-parameter collaborative optimization for electro-optical imaging systems

Existing optimization approaches for infrared electro-optical imaging systems largely rely on analytical models based on linear approximations, while non-ideal imaging factors are rarely incorporated into a unified system-level design framework, limiting the reliability in performance prediction. To address the requirements for detection and recognition of infrared targets in complex scenes, a full-chain physical modeling framework for infrared electro-optical imaging is established, from radiative input through photoelectric conversion to grayscale image output, enabling a unified characterization of the nonlinear coupling among spectral response, spatial transfer, sampling, and noise. Based on this framework, simulated image contrast is adopted as the objective function. Key parameters, including F-number, pixel size, integration capacitance, veiling glare index, and the root mean square (RMS) radius of the aberration spot, are selected as design variables by analyzing the influence of system component parameters on the multidimensional propagation chain. Structural parameters and non-ideal imaging factors are thereby coupled within a unified performance evaluation space, enabling a cross-level multi-parameter collaborative optimization model that integrates performance optimization with physical constraints. A hybrid strategy combining global search via genetic algorithms and local search is employed to efficiently solve the resulting nonlinear optimization problem, and the feasible ranges of non-ideal imaging parameters are further inversely derived from system performance thresholds. Simulation results under land-background and sea-clutter scenes demonstrate stable convergence and significant contrast improvement, indicating strong cross-scene adaptability. The proposed framework provides a systematic approach for the digital collaborative design of high-performance infrared imaging systems in complex environments.