Absorption images as obtained by imaging riometers such as IRIS are usually created by interpolating between absorption values for individual beams. For IRIS, the locations of the beam centres serve as grid points for subsequent linear interpolation. Although generally producing good results, the fact that the actual shape of the imaging beams is not considered, potentially introduces errors and can lead to misinterpretations. In this paper, two alternative interpolation methods are introduced. Method A is based on measuring the similarity between simulated reception of individual point sources and actually received data. Method B uses a mathematical model of the sky brightness distribution parametrised by the received data. All interpolation methods are applied to power data, as opposed to absorption data, in order to avoid any errors that might be introduced by intermediate processing steps, especially QDC (quiet-day curve) generation. We apply all methods to synthetically generated test data as well as to three exemplary real datasets which are also compared to a calculated sky brightness distribution obtained from a skymap.