SSP Project Summary:

Recovery of depth information from optical microscope images by constrained deconvolution

There was lots of maths in the original proposal. What follows is a potted summary extracted from an article submitted to the EPCC newsletter.

Optical microscopes produce poorer images than either electron or confocal microscopes, but they are far cheaper. Digital methods can be used to enhance light microscope images. For example, deconvolution can improve the resolving power, and depth information can be extracted from blur. In optical sectioning, a series of images are taken of a specimen with the optical system focussed at different depths perpendicular to the focal plane of the system. Different parts of the specimen are in focus in each image, but these can be obscured or made ambiguous by unfocussed information from other parts of the specimen. The blur can be represented as a linear convolution of the specimen. Code exists that solves the relevant equations to improve the quality of optical microscope images. However, the processing takes several days to reach convergence on available workstations.

It is hoped that the application of parallel processing to this problem will help make practical the use of optical microscopes, even where high quality images are required, by significantly reducing the run-time needed for the image improvement.

Michael Perkins worked on this project.

Compressed PostScript of the project's final report is available here (424 kbytes) .

Webpage maintained by mario@epcc.ed.ac.uk