Research on detection and correction of defective pixels of LASIS

Due to the imperfections in producing, a finite number of pixels in an array will be defective. Regarding a Fourier transform Imaging spectrometer, the existence of defective pixels will not only affect the quality of the image, but also cause interferogram extraction error, and then result in a distortion of the reconstructed spectrum. So the defective pixels must be accurately distinguished and eliminated by data processing. In this paper, according to the characteristic of the Fourier transform Imaging spectrometer, a distinguishing and eliminating method is carried out. The Fourier transform Imaging spectrometer is illuminated by a uniform light. The data is fitted in the spatial dimension; the error between the actual data and the fitted data is computed and divided by the standard deviation. By choosing a proper threshold value, the cold, hot and non-saturated pixels can be effectively distinguished. Single defective pixels can be effectively corrected by spatial dimension interpolating; for clustered defective pixels, spatial dimension interpolating and interference dimension fitting are taken and the result is averaged. The experimental result proves that this method is effective and also efficient both for uniform light illuminated data and push broom data.