MI-based correction of intensity inhomogeneity using singularity function analysis

We proposed a new approach for correcting intensity nonuniformity (intensity inhomogeneity, bias field, or shading), that hampers the use of automatic image processing techniques. The intensity nonuniformity is perceived as a smooth intensity variation across the image. The proposed approach is on the basis of singularity function analysis, and mutual information is criterion of stopping iterative process. The rationale is that the low frequency component containing bias field is removed, followed by reconstruction from its high frequency component not containing bias field. Then the estimated bias field and the corrected image can be obtained