Nonuniformity correction for internal radiation in uncooled infrared imaging based on gradient-weighted guided image filtering

Due to the absence of a cold screen, the performance of uncooled infrared focal plane array imaging is significantly affected by internal radiation caused by optical systems or cavities during long-term operation. Previous nonuniformity correction methods have struggled to effectively eliminate fixed-pattern noise resulting from internal radiation. The paper explores an internal radiation correction method termed GWGF-IR, which utilizes a gradient-weighted guided image filter to correct spatially continuous fixed-pattern noise (FPN) in uncooled infrared focal plane array imaging systems. This method incorporates gradient weight factors to constrain the weights of different pixels, forming the weighted quadratic cost function, resulting in a gradient-preserving smoothing filter kernel designed for effectively extracting internal radiation noise gradients in infrared imaging. Furthermore, the paper proposes a self-adaptive estimation method for determining gradient weight factors and regularization parameters. The GWGF-IR algorithm demonstrates adaptive nonuniformity correction of internal radiation across varying intensities and scene images, outperforming current typical methods in subjective visual effects and objective evaluation indicators. The processing time on a 324 × 256 image can reach 24.8 ms, which is 45.7% of the time required by the suboptimal algorithm; The PSNR index can outperform the suboptimal algorithm by 41%, and the MMSIM index can outperform the suboptimal algorithm by 0.23%.