Most discrete image processing algorithms are linear: the elements of the processed image are formed as a linear combination of the elements of the original image. The widespread use of linear operations is explained by the fact that they are easier to perform than nonlinear ones. However, when processing large-sized arrays, it is often not possible to perform even linear operations if there are no efficient calculation algorithms. This chapter discusses indirect computational methods that rely on unitary transformations and allow processing to be performed more efficiently than using traditional methods. It also describes recursive two-dimensional filtering methods.
Comments