Fraunhofer-Gesellschaft

Publica

Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

An optimal non-orthogonal separation of the anisotropic gaussian convolution filter

 
: Lampert, C.H.; Wirjadi, O.

:
Fulltext urn:nbn:de:0011-n-464046 (2.7 MByte PDF)
MD5 Fingerprint: c961cf3181d46301fd8631b24be83c22
Created on: 13.09.2006


Kaiserslautern: Fraunhofer ITWM, 2005, 38 pp.
ITWM-Berichte, 82
English
Report, Electronic Publication
Fraunhofer ITWM ()

[]
Vorwort S.3-4
1 Introduction S.5-6
- 1.1 Anisotropic Gaussian Filtering S.5-6
- 1.2 Our Contribution S.7
2 The Gaussian Convolution Integral S.7-9
- 2.1 Factorization of the Gaussian S.7-9
- 2.2 Separating the Convolution Integral S.9
3 Symmetric Factorizations of - S.10-12
- 3.1 Singular Value Decomposition S.10
- 3.2 Triangular Factorization of Cholesky Type S.10-11
- 3.3 Calculation of the Triangular Factorization S.11-12
- 3.4 Geometrical Interpretation S.12
4 Anisotropic Gaussian Filtering in Image and Signal Processing S.13-14
- 4.1 Parameterizations for Image Processing in 2D S.13-14
- 4.2 Parameterizations for Image Processing in 3D S.14
5 Discretization & Implementation S.15-17
- 5.1 Discrete Directional Convolution Operator S.15
- 5.2 Interpolation S.15-16
- 5.3 Finite and Infinite Impulse Response Filtering S.16-17
- 5.4 Implementation S.17
6 Results S.18-22
- 6.1 Accuracy S.18
- 6.2 Speed S.18-22
7 Discussion S.23
- 7.1 Accuracy S.23
- 7.2 Speed S.23
8 Applications S.24-26
- 8.1 Processing of Fibrous Data S.24-25
- 8.2 Spatiotemporal smoothing of Video Sequences S.25-26
Conclusion S.27
References S.28-29
Published reports of the Fraunhofer ITWM S.30-38