Deconvolution of 3D Fluorescence Microscopy Images by Combining the Filtered Gerchberg-Papoulis and Richardson-Lucy Algorithms

Moacir P. Ponti JuniorNelson D.A. MascarenhasMarcelo R. ZorzanClaudio A.T. SuazoMurillo R.P. Homem

The problem of deconvolution in fluorescence microscopy deals at the same time with the diffraction limit that cuts off some of the frequencies, blurring the image, and photon noise, that corrupts the image by inserting elements that are not present in the real object and also distorting the contrast. This distortions hampers the possibility of using the 3D images for recognition and analysis applications. In addition, the algorithms developed, in general, assume absence of noise or a white additive noise. This work presents an approach to deconvolve the images and deal with the noise present in real images. The Gerchberg-Papoulis algorithm and a smoothing operator were combined with the Richardson-Lucy iterative algorithm. The results of the method for simulated data are compared with the ones obtained by the original algorithm. The method improved the results by obtaining higher signal-to-noise ratio and quality index values, performing a better band extrapolation.

Caso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato:
     Mantida por: