A fuzzy inference method for image fusion/refinement of CT images from incomplete data

The quality of computed-tomography (CT) images deteriorates when images are reconstructed from incomplete data. This work makes use of the a priori knowledge inherent in the membership functions and the logical rules of a fuzzy inference system (FIS) to compensate for the missing data. It is shown that a fuzzy inference system can be used to improve the quality of reconstructed CT images, particularly when the images are reconstructed from incomplete data. It is proposed to reconstruct a coarser image for which the data is over-complete, and use the histograms of this image and that of the original finer image to generate the membership functions required in FIS. The two images are then fused, with the aid of logical rules based on the knowledge that the two images posses the same distinct attributes (pixel values). In order to avoid the difference in spatial resolution between the original fine image and the reconstructed coarse image, a modified FIS method is introduced to refine the fine image. Results are presented, showing visually and quantitatively that this FIS refinement process improves the quality of the original fine image.