Image restoration of biological pictures can he achieved by convolution of the image received with a spread function simulating a corrective imaging which cancels out the degradation incurred during actual imaging in the microscope. A method is described by which a substantial fraction of the restoration can be achieved by sampling only six neighbouring locations for each point processed, thus speeding up substantially the restoration process without seriously affecting the quality of the results. The effective resolution of the system can be approximately doubled. Curves are given for the selection of the two parameters controlling the processing, according to the charecteristics of the imaging device to be improved.