Correction Method for Keystone Distortion and Smile Distortion Based on Subpixel Transformation

This paper proposes a sub-pixel transformation method to accurately correct Keystone and Smile distortions in astronomical spectra from instruments like FASOT and LAM-OST. Peak detection with quartic polynomial fitting restores the fiber output envelope, while valley detection reveals wavelength shifts from effects like the Doppler effect. Pixel deviations remain within 0.50, with most below 0.10, preserving spectral integrity. Applied to LAMOST spectra, the method reduces fiber interference, narrows projection bandwidth, and increases peak spacing. This approach ensures precise calibration without introducing distortion, improving spectral analysis accuracy, and imaging quality.