Image Restoration Algorithm Based on Regularization and Adaptation

We propose to add a special summand (stabilizer) to the original Tikhonov regularization algorithm; this regularizer includes a specially adapted function to the solution characteristics for the problem of image restoration. This approach to approximation of non-smooth functions based on our new technique for choosing interpolation points. As a result, the approximate solutions have better accuracy and images become more deblured. Moreover, it becomes possible to keep small objects and contours in complex scenes by incorporation of background knowledge about their location or structure into the regularization procedure.