Abstract
The problem of enhancement of tomographic image quality and the related problems caused by the conflict between the resolution and stability of the restoration method are considered. The Tikhonov regularization method is proposed to solve these problems. The method allows the ratio of accidental error and systematic inaccuracy of restoration to be varied by setting the regularization parameter and regularization order, which allows an optimal proportion between them to be found. A priori information in the form of constraints is applied to the reconstructed images. The use of such information decreases the time cost of reconstruction and enhances the quality of the reconstructed images, including improvement in image contrast. The results of computer modelling are given. The relative mean square deviation between the reconstructed image and the exact image, and the mean smoothness of the reconstructed image are used as characteristics of accuracy of restoration. A comparison of the reconstruction quality of the regularized solution for different regularization parameters and regularization orders in the presence of uniformly distributed 5% noise applied to the projection data is performed.