next up previous
Next: Image Recognition Up: Image Analysis Previous: The SECB Method

Computer Implementation of the SECB Image Deblurring Method

Marjorie McClain, ACMD

In connection with the image deblurring project of A. Carasso, software has been developed to deblur a PET image for several trial values of the SECB parameter and then display the deblurred images. Other deblurring methods require intensive computation, whereas the SECB method offers the potential for quite reasonable computing times on personal computers, making this method attractive for use in medical settings without advanced computing facilities.

The present Unix implementation of the SECB deblurring software consists of two parts: (1) a Fortran program that reads in a blurred image, specified as a matrix of bytes representing gray levels, and performs the deblurring calculations for the desired values of the SECB parameter, and (2) a PV-Wave program that displays the deblurred images. Work this year has involved restructuring the Fortran program to reduce its memory requirements, resulting in a program that will run under MS-DOS on a PC. This program can deblur a 128x128-byte image for ten parameter settings in about 1.5 minutes on a 486 PC, demonstrating the feasibility of developing the SECB method for use on personal computers. Future work will involve combining the computation and display programs into one program, with graphics appropriate for a PC.