The proposed research is directed towards the development, refinement, evaluation and biological application of deconvolution algorithms applied to image enhancement of three-dimensional images obtained from the light microscope. Specifically, it is proposed to (1) develop maximum-likelihood deconvolution algorithms based on a linear parametric object model. These are anticipated to have broad applicability to both wide-field and optical sectioning microscopies. (2) to develop methods combining pattern recognition and image restoration for imposing structural constraints on maximum-likelihood data. This should improve resolution and enable the restoration to adapt to feature in the data. (3) develop a suite of evaluation tools based on computer simulation, custom fabricated test objects and biological specimens. These tools will be used to investigate the behavior of the algorithms under a variety of conditions so as to optimize these algorithms and gain a comprehensive understanding of their performance.
