Biomedical Imaging Group
EPFL   Student Projects: François Aguet
English only    BIG > Teaching > Student Projects > François Aguet
Home page
Tutorials and Reviews
Download Algorithms
bulletStudent projects
Three-dimensional surface recovery from microscopy images using object model-based deconvolution

François Aguet
Section Informatique, EPFL

Diploma project
March 2004

We present a method for the recovery of the three-dimensional surface of a thick specimen in brightfield microscopy. Using a stack of images of the object at different focal distances, we reconstruct its surface height map and surface texture. Clearly, to be useful the object must correspond to the model’s assumptions. Once we have obtained the surface height map, it can be used to extract the surface texture, which will be a single, completely in-focus image. This approach can be considered as a parametric deconvolution (by imposing the object model). As such, our method can be applied as an extended depth of focus algorithm. The method incorporates the diffraction-limited point spread function model of a microscope. We present the validation of the algorithm using simulation experiments. Finally, we also show results for a real-world experimental setup.