TU Darmstadt / ULB / TUbiblio

Lazy Nonlinear Diffusion Parameter Estimation

Thürck, Daniel ; Kuijper, Arjan (2013):
Lazy Nonlinear Diffusion Parameter Estimation.
In: Lecture Notes in Computer Science (LNCS); 8156, pp. 211-220, Springer, Berlin, Heidelberg, New York, Image Analysis and Processing - ICIAP 2013. Proceedings Part I, DOI: 10.1007/978-3-642-41181-6₂₂,
[Conference or Workshop Item]

Abstract

Perona-Malik diffusion is a well-known type of nonlinear diffusion that can be used for image segmentation and denoising. The process itself needs an parameter k to decide which edges will be retained and which can be blurred and a stopping time tS. Although there have been investigations on how to set these parameters, especially for regularized diffusion models, as well as different criteria for the optimal stopping time have been suggested, there is yet no quick and conclusive way to estimate both parameters - or to reduce the search space at least. In this paper, we show that Gaussian noise characteristics of an image and the diffusion parameters for an optimal optical result can be estimated based on the image histogram. We demonstrate the effectiveness of lazy learning in this area and develop a custom feature weighting algorithm.

Item Type: Conference or Workshop Item
Erschienen: 2013
Creators: Thürck, Daniel ; Kuijper, Arjan
Title: Lazy Nonlinear Diffusion Parameter Estimation
Language: English
Abstract:

Perona-Malik diffusion is a well-known type of nonlinear diffusion that can be used for image segmentation and denoising. The process itself needs an parameter k to decide which edges will be retained and which can be blurred and a stopping time tS. Although there have been investigations on how to set these parameters, especially for regularized diffusion models, as well as different criteria for the optimal stopping time have been suggested, there is yet no quick and conclusive way to estimate both parameters - or to reduce the search space at least. In this paper, we show that Gaussian noise characteristics of an image and the diffusion parameters for an optimal optical result can be estimated based on the image histogram. We demonstrate the effectiveness of lazy learning in this area and develop a custom feature weighting algorithm.

Series Name: Lecture Notes in Computer Science (LNCS); 8156
Publisher: Springer, Berlin, Heidelberg, New York
Uncontrolled Keywords: Image analysis, Image processing, Machine learning, Parameter ranges
Divisions: 20 Department of Computer Science
20 Department of Computer Science > Interactive Graphics Systems
Event Title: Image Analysis and Processing - ICIAP 2013. Proceedings Part I
Date Deposited: 12 Nov 2018 11:16
DOI: 10.1007/978-3-642-41181-6₂₂
Export:
Suche nach Titel in: TUfind oder in Google
Send an inquiry Send an inquiry

Options (only for editors)
Show editorial Details Show editorial Details