Inverse filtering of optical images

The quality of images is limited by the performance of the optical system used. The imperfections of the optical system cause distortion of the image. If the distortion is known it can be (partly) compensated. This procedure is called inverse filtering. The problem is, however, ill-posed, which means that the measurement noise is amplified by the inverse filtering process. Suppression of the noise causes bias in the reconstruction. A tradeoff has to be found between the noisy and biased estimates. In this paper, the reconstruction of images will be investigated, assuming that the distortion of the optical system is known. An algorithm is introduced to estimate the optimal level of noise suppression of the two-dimensional inverse filter