Thanks to high angular resolution abilities, the hypertelescope concept forms a promising way to develop new astronomical instruments. However, its feasibility have to be experimentally demonstrated and especially a solution to the cophasing problem has to be found. This is the aim of this PHD work. In the XLIM laboratory, we have developed a fibered hypertelescope prototype called Temporal HyperTelescope (THT). The concept of this height-telescope-instrument is detailed in the first chapter of this document. A study of the instrument defaults has been carried out to evaluate its theoretical imaging abilities. In a second step, a cophasing system has been implemented. It is based on a joint use of a genetic algorithm and the phase diversity technique. Its efficiency has been experimentally validated in laboratory thanks to the acquisition of a binary star image showing a 9.1 magnitude unbalance and requiring a 3-nanometer-optical-path-lengh-resolution. Then, we have tested this method in the photon counting regime and the experimental results demonstrate that the instrument is able to acquire images even in these extreme running conditions. Finally, the last chapter of this document gives few points to be improved and suggests a draft of space based instrument which could be realized in the next decade.
