A Multigrid Method for Total Variation-Based Image Denoising
C. R. Vogel · Birkhäuser Boston eBooks · 1995
Byimage reconstruction we mean obtaining the solution u of an operator equation of the form (1.1) $$Au = z,$$ where A is typically (but not necessarily) linear, and the dataz is assumed to be contaminated by error. Two important special cases are: (i) Denoising, or noise removal, whereAu =u; and (ii)Deblurring, whereA is a Fredholm first kind integral operator (1.2) $$(Au)(x) = \int_\Omega k (x,y)u(y)dy,x \in \Omega $$ which models “blurring” processes involved in image formation. See [9] for details.