Home

Edge-preserving and scale-dependent properties of total variation regularization


Author(s) : Tony F. Chan David M. Strong, 
Publisher : N/A
Publication Date : 2000
ISSN : N/A
Abstract : We give and prove two new and fundamental properties of total variation minimizing function regularization (TV Regularization): 1) edge locations of function (e.g. image) features tend to be preserved, and under certain conditions, are preserved exactly; 2) intensity change experienced by individual features is inversely proportional to the scale of each feature. More generally, we describe both qualitatively and quantitatively the exact eects of TV Regularization in R 1,