Photometric parameter inversion and nonlinearity characterization of infrared focal plane arrays based on flat-field...

In infrared time-domain astronomy and high-precision radiometric measurements, the photometric accuracy of infrared focal plane arrays (IRFPAs) is limited by the coupled effects of non-ideal factors such as gain bias, response nonlinearity, and spatial correlation. The conventional photon transfer curve (PTC) method cannot adequately characterize variance redistribution and nonlinear modulation, which limits the accuracy of key parameter extraction. To address this issue, this work proposes a physics-driven parameter inversion method based on flat-field data. A unified forward simulation model is constructed, in which photon shot-noise propagation, inter-pixel spatial correlation, and signal-dependent response nonlinearity are incorporated into a single framework. Under a unified statistical convention, statistical features including the central variance, the local covariance sum, and the spatial-correlation structure are jointly extracted, and a closed-loop parameter inversion procedure is established to enable the joint identification of gain, nonlinearity, and spatial-correlation parameters. Experimental results from a short-wave infrared (SWIR) detector show that the proposed method reproduces the measured PTC curve with a residual below 2%, captures the key spatial statistical features well, effectively suppresses systematic bias in photometry on real star-field images, and limits the relative prediction error of the core variance statistics to within 3.2% in cross-date prediction tests. These results indicate that the proposed framework provides a feasible technical approach for high-precision photometric characterization and ground calibration of non-ideal infrared detectors.