An image of a sea bottom observed through a rough sea surface is randomly distorted by the refraction of light at the water-air interface. The negative effect of waves on bottom visibility can be decreased by 'accumulating" image frames. Realistically the possibility of image accumulation is very limited. As a rule an observer has to deal with random images distorted by the rapidly changing sea surface. Thus a question arises: Is it possible to retrieve the "real" image of an object from its distorted image, if an image of the exact water surface distorting the bottom image is made simultaneously? The present paper addresses the correction of a distorted image of a bottom object, when the water surface relief or the spatial distribution of surface slopes is precisely known. Additionally a correction algorithm is given for rapidly processing two-dimensional images, the potential for determining the "instantaneous" spatial distribution of rough sea surface slopes from its instantaneous image is investigated, and finally an analysis is made relating errors in determining sea slopes from a single image to the associated quality of image correction for an object observed through a changing sea surface.© (2007) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.