Incremental restoration of SPECT images

Presents a new method that incrementally restores the resolution of an image, using VCD (variable conductance diffusion) filters to suppress noise before restoring resolution. It is applied post-reconstruction to SPECT image sets which have known Gaussian blur. At each step, a VCD filter is applied to the difference between the unregularized image restored to the previous resolution and that restored to the resolution at the current step. The filtered difference is then added to the result from the previous step, which serves as a base image for the current step. The VCD filter is determined from the information given by the sum of the base image and the filtered difference. When this method is implemented in one step, better resolution than that from Butterworth regularization, and smoother edges than those from direct VCD regularization are observed. This demonstrates that the idea of forming the filter from one set and applying it to the other works well. A five-step incremental restoration performs even better than one-step, which shows the advantage of restoring an image incrementally.<
>