On restoration of hard-contrasted images

A method is proposed which computes the PSFs (point spread functions) in the space domain after reconstructing recognized coarse and, by preference, hard-contrast structures in an image, either by making a facsimile of them or by slicing the original; the reconstructed sharp shapes and the corresponding blurred image parts are then used as the knowns for the solution of the convolution equation. The gray values assigned by scanning devices do not satisfy the linearity assumption needed for the application of this procedure-the difference being visible mainly in hard-contrasted image parts.<>