Dies ist eine Übersichtsseite mit Metadaten zu dieser wissenschaftlichen Arbeit. Der vollständige Artikel ist beim Verlag verfügbar.
Edge-preserving and scale-dependent properties of total variation regularization
597
Zitationen
2
Autoren
2003
Jahr
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 , R 2 and R 3 . We give and prove exact analytic solutions to the nonlinear TV Regularization problem for simple but important cases, which can be used to better understand the eects of TV Regularization for more general cases. The formulae we give describe the eect of TV Regularization when applied to noise-contaminated radially symmetric image features. These formulae also accurately predict the eects of TV Regularization when it is applied to more general functions. Our results help explain how and why TV...
Ähnliche Arbeiten
ImageNet classification with deep convolutional neural networks
2017 · 75.672 Zit.
Unpaired Image-to-Image Translation Using Cycle-Consistent Adversarial Networks
2017 · 21.622 Zit.
Nonlinear total variation based noise removal algorithms
1992 · 15.555 Zit.
High-Resolution Image Synthesis with Latent Diffusion Models
2022 · 12.713 Zit.
Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network
2017 · 12.200 Zit.