Abstract: The physical limitations of detector systems cause stripe non-uniformity in scanning infrared imaging, severely degrading the image quality through stripe noise. However, existing non-uniformity correction algorithms struggle to balance noise suppression, detail preservation, and real-time performance, thereby limiting their use in spectral imaging and signal processing. To address the challenge of stripe noise, we propose an innovative wavelet-transform-based method to minimize the fractional total variational function. Based on the wavelet transform, the pixel neighborhood gradient serves as an adaptive parameter of the entire algorithm. By leveraging the relationship between stripe noise characteristics and wavelet sub-band coefficients, the proposed method efficiently removes stripe noise while accurately reconstructing image details with low computational cost. Comprehensive experiments on authentic data demonstrate that the proposed method outperforms several existing classical algorithms in both quantitative and qualitative evaluations.