A Shutter-Less Nonuniformity Correction Algorithm Based on Noise Response Model for TEC-Less Uncooled Infrared Sensors

This article proposed a nonuniform correction algorithm based on the noise response model without shutter and thermoelectric cooler (TEC). Specifically, the nonuniform noise exhibited by the imager at different temperatures is extracted and decomposed into three typical noises: “halo” noise, stripe noise, and high-frequency fixed pattern noise (HF-FPN). Three kinds of nonuniform noise are corrected by grading: 1) for “halo” noise, a correction algorithm based on a 2-D Gaussian surface diffusion model is proposed, which realizes the real-time correction of “halo” nonuniformity by six parameters and 2) for stripe nonuniformity and HF-FPN, a local optimization algorithm based on a noise response template is proposed. The experiments show that the nonuniform noise model and the correction algorithm proposed in this article are better than the current widely used correction algorithm in the four indicators: Var (≤3000), NUES (≤60), RI (≤0.007), and PSNR (≥95 dB) under the long-time stable state (more than 120 min), and the visual effect is more uniform. In the process of rapid temperature change at 10 °C/min, it also shows the excellent correction performance in subjective visual and objective indicators. At the same time, the resource and complexity of the algorithm are only equivalent to the correction algorithm of multipoint calibration, which can be implemented on embedded platforms such as FPGA.