An artifact reduction method of the penumbral images by using kernel principal component analysis

摘要

In this paper, we proposed a new method to obtain a clear reconstructed image from noisy penumbral images. The reconstructed image can be obtained from the penumbral image by deconvolution. Usually, experimentally obtained penumbral images contain noise and their signal-to-noise ratio (S/N) is low. It becomes difficult to obtain a reconstructed image and the reconstructed image contains artifacts. We use kernel principal component analysis (KPCA) to remove the artifact from the reconstructed image. The efficacy of the proposed method is demonstrated by computer simulations.