The developed system is a tool for high aperture imaging of the fundus. The obtained high resolution images preserve the topology
of the blood vessels. The system is based on mosaicking a series of distinct low aperture fragments in order to obtain a high
aperture image. Mosaicking is implemented by a neural network with stubborn learning taking into account the importance of
the information of particular features. In mosaicking, the aberrations of the third order are partly compensated.