Zeng, QJ; Qin, HL; Yan, X; Zhou, HX (2020). Fourier Spectrum Guidance for Stripe Noise Removal in Thermal Infrared Imagery. IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, 17(6), 1072-1076.

Thermal infrared (TIR) imaging has been an indispensable tool in surveillance and remote sensing fields due to the characteristic of this spectrum that enables the sensing system to detect relatively warm targets, especially in low-light conditions. However, the acquired TIR images often suffer from observable stripe noise, which reduces the target detectability to some extent. To remove the noise and keep the image details, this letter proposes a novel method that combines the spectral processing technology with the image-guidance mechanism. Specifically, the frequency band contaminated by stripe noise is corrected with the corresponding Fourier coefficients of a guided image, which can be estimated by existing smoothing methods. Various experiments on the simulated and real TIR images show high performance and efficiency of the proposed method. In addition, in the application of small target detection, it is demonstrated that local contrast between the target and its background is well maintained and the signal-to-clutter ratio is increased when our method is performed.