Abstract | ||
---|---|---|
In this paper we derive a unified framework for the taut-string algorithm and regularization with G-norm data fit. The G-norm data fit criterion (popularized in image processing by Y. Meyer) has been paid considerable interest in regularization models for pattern recognition. The first numerical work based on G-norm data fit has been proposed by Osher and Vese. The taut-string algorithm has been developed in statistics (Mammen and van de Geer and Davies and Kovac) for denoising of one dimensional sample data of a discontinuous function. Recently Hinterberger et al. proposed an extension of the taut-string algorithm to higher dimensional data by introducing the concept of tube methods. Here we highlight common features between regularization programs with a G-norm data fit term and taut-string algorithms (respectively tube methods). This links the areas of statistics, regularization theory, and image processing. |
Year | DOI | Venue |
---|---|---|
2005 | 10.1007/s10851-005-6462-1 | Journal of Mathematical Imaging and Vision |
Keywords | Field | DocType |
regularization model,tube method,g-norm data fit,higher dimensional data,regularization program,regularization programs,regularization theory,taut-string,taut-string algorithm,g-norm,denoising,g-norm data,image processing,dimensional sample data,pattern recognition,data fitting | Noise reduction,Continuous function,Mathematical optimization,Image processing,Algorithm,Regularization (mathematics),Regularization theory,Mathematics,Regularization perspectives on support vector machines | Journal |
Volume | Issue | ISSN |
23 | 2 | 0924-9907 |
Citations | PageRank | References |
4 | 0.74 | 4 |
Authors | ||
1 |
Name | Order | Citations | PageRank |
---|---|---|---|
Otmar Scherzer | 1 | 346 | 52.10 |