Dual-Gain SWIR High Dynamic Range Imaging Based on Irradiance Normalization and Continuous Logarithmic Tone Mapping

Abstract

In industrial process monitoring, short-wave infrared (SWIR) imaging often encounters extreme luminance variations that exceed the dynamic range of conventional SWIR cameras, requiring simultaneous preservation of dark-region details and bright-region information. For dual-gain SWIR cameras, the high conversion gain (HCG) mode enhances dark-region visibility, whereas the low conversion gain (LCG) mode preserves bright-region details. Neither single-gain mode can adequately retain both dark-region and bright-region details. To address this issue, a dual-gain SWIR high dynamic range (HDR) imaging method based on irradiance normalization and constrained continuous logarithmic mapping is proposed. Experimental results from laboratory tests and field validations verify that the proposed method achieves smoother grayscale transitions, improved dark-region visibility, and effective suppression of highlight overexposure. Local profiles, ablation results, and regional metrics further validate the proposed method. Quantitative analysis of outdoor building scenes yields performance improvements of 12.13%, 6.11%, and 20.13% in information entropy, standard deviation, and average gradient, respectively, compared with the linear hard splicing method.