The DSNR (discriminative signal-to-noise ratio) optimization approach is extended to include more than one template. The optimal DSNR template matching filter can be designed to elicit any desired response for each training template image while optimizing the DSNR criterion. The approach used considers additive noise as a parameter and leads to a very general formulation, of which many previous approaches are special cases. In the case of a single training image, this formulation reverts to the Wiener restoration filter or the template-similar expansion approach.
展开▼
