Improvement of the LLS and MAP deconvolution algorithms by automatic determination of optimal regularization parameters and pre-filtering of original data

Bruno Colicchio (1) , Olivier Haeberlé (1) , C. Xu (1) , Alain Dieterlen (1) , Georges Jung (2)

1 MIPS - Modélisation, Intelligence, Processus et Système
2 Département d'hématologie

We show that automatic determination of regularization threshold and pre-filtering of 3-D fluorescence microscopic images improves the stability of deconvolution results when using the Linear Least squares Solution or the Maximum a Posteriori method. Doing so, the choice of the regularization parameter much less depends on a priori knowledge of the specimen or skills of the operator. This increases the reliability and repeatability of quantitative measurements on deconvolved images.

Mots clés: Deconvolution; Pre-filtering; 3-D microscopy

Domaines: Traitement du signal et de l'image [eess.SP]; Traitement du signal et de l'image [eess.SP]

Bruno Colicchio, Olivier Haeberlé, C. Xu, Alain Dieterlen, Georges Jung. Improvement of the LLS and MAP deconvolution algorithms by automatic determination of optimal regularization parameters and pre-filtering of original data. Optics Communications, 2013, 244 (1-6), pp.37-49. ⟨10.1016/j.optcom.2004.08.039⟩. ⟨hal-00919927⟩