Results

 Fig. 4 shows one of the objects used in the simulation, a coloured cluster of 27 stars. The object's intensity distribution was convolved with the interference function of the interferometer and then multiplied with the diffraction function, both functions being converted to intensities. The optical path differences were found, using 3 wavelengths, with initial wavefront errors of the order of $10 \mu m$. The Strehl ratio, i.e. the peak intensity in the image of a point source, relative to the case of perfect phasing, gives an idea of the phasing performance. It reaches 30% with 2200 photons detected during the phasing cycle and 80% with $18 \cdot 10^6$ photons.

The successive frames of Fig. 6 from left to right depict changes in the full combined image during the phasing steps for the same object.

The field and object size restriction with densified-pupil imaging often implies that the object is unresolved or little resolved by the smallest baselines. Although the simplified routines which we used in the simulations did not include all the features which appeared necessary, the resulting imaging quality approached the theoretical limits on the source shapes used, typical of resolved stellar and galactic objects. Even better results should therefore be expected with future refined versions of the algorithm.

 

Figure 6: Simulated phasing sequence on galaxy M51, in presence of photon noise. The galaxy is shown in the top-left frame, followed by the interference function of the interferometer and the convolution of the object with the interference function. The object is unresolved by the smallest baselines, its components have different colours and phasing is achieved with 250,000 photons. Successive frames depict changes in the full combined image during the phasing steps. The last frame, at bottom right, is the final step of the phasing process, where a Strehl ratio of 0.80 was achieved.