Correction of fixed pattern background and restoration of JPEG compressed CCD images

Abstract

Summary form only given; substantially as follows. The present paper addresses the problem of the removal of the sensor background patterns, dark current and responsivity, from CCD images, when the uncorrected image was transmitted through a JPEG like block transform coding system. The work is of particular interest for imaging systems which operate under severe hardware restrictions, and require high accuracy, e.g. deep space cameras. The complexity of the problem comes from the aliasing of the image signal and CCD background patterns during the quantization in the transformed domain. The authors investigated several solutions and selected the optimal one based on three objectives: the radiometric accuracy, the visual quality, and the computational complexity. The solution selected for the background pattern removal and image restoration uses a combination of different methods: correction in space domain and iterative regularization in both space and DCT domain.