Adaptive Bad Pixel Correction Method for Interference-Modulated Images Based on Weighted Least Squares Support Vector Machines (WLS-SVM).

Temporally and spatially modulated Fourier transform imaging spectrometers (TSMFTISs) can obtain images and interference information of targets during the data acquisition process for remote sensing. Temporally and spatially modulated Fourier transform imaging spectrometers play an important role in target classification and identification, as the spectrum information of targets can be reconstructed with the theory of Fourier transform spectroscopy. However, the defect pixels absent in the planar array charge-coupled device used in imaging spectrometers have a significant impact on the accuracy of target spectral recovery information, so the preprocessing of bad pixels in remote sensing interference images is indispensable to data processing in TSMFTIS. An adaptive defect pixel correction method based on the weighted least squares support vector machine is introduced in this paper. The principle of TSMFTIS is presented to state the specialty of bad pixels and discuss the limitations of the traditional defect pixel method. Simulations based on the conventional method and the proposed method are performed to obtain bad pixel correction results for TSMFTIS. The algorithm presented in this paper is more efficient and robust. An application of the proposed method is employed.