In fluorescence microscopy, low light intensities must often be used for imaging to avoid undesired effects such as photobleaching, or photochemical alterations of the sample. Noise suppression (due to the low light intensities) in the acquired data is thus a critical issue in that context. In practice, one can perform several acquisitions in time, in order to obtain several noisy versions of the same image. A less noisy result can then be obtained by simple averaging or by unknown techniques of the associated microscopy software. The goal of this project is to derive and apply an efficient denoising technique for this problem. In our case, the denoising algorithm takes the different noisy acquisitions as an input, and returns the denoised result. The denoised solution must be close (up to the noise distortion) to the measurements, and has to satisfy some prior knowledge on the data. Our interest is also to pioneer --- if time permits --- problem formulations that are well adapted to the noise model. The intuition is that such formulations are particularly interesting and can substantially improve the quality of the results when considering several acquisitions. Applications of the method to real biomedical data will be considered. The results can be compared to the standard software solution that is normally used in that case. Requisites : courses in signal/image processing, interest for algorithmic methods and general knowledge in programming (MATLAB and/or Java).