Joint denoising-deconvolution approach for fluorescence microscopy
Suman Maji, Catherine Dargemont, Jean Salamero, Jérôme Boulanger · 2016
Image denoising and deconvolution are well known techniques applied to wide-field and confocal microscopy in order to restore images. The methods however suffer from their own drawbacks with denoising potentially resulting in smoothed images while deconvolution giving unpleasant artifacts due to ill-posedness of the problem. In this paper, instead of trying to tackle the problem of low signal-to-noise ratio (SNR) images in microscopy imaging, through regularization or using denoising as a pre-processing step to deconvolution, we propose to evaluate the interest of restoring the unknown image through a joint denoising and deconvolution program applied simultaneously in an iterative framework. We propose to minimize four different energy functional arising through different combination of regularizing terms and evaluate the gain obtained by this approach. Simultaneously addressing the two problems, we demonstrate on synthetic images that a join approach is able to provide superior qualitative and quantitative results, both in the case of low and high noise scenarios. We further illustrate the performances of this approach on a real image.