Scale-Space Anisotropic Total Variation for Limited Angle Tomography

Huang Y, Taubmann O, Huang X, Haase V, Lauritsch G, Maier A (2018)

Publication Language: English

Publication Type: Journal article, Original article

Publication year: 2018


Book Volume: 2

Pages Range: 307-314

Article Number: 1

Journal Issue: 4


DOI: 10.1109/TRPMS.2018.2824400

Open Access Link:


This paper addresses streak reduction in limited angle tomography. Although the iterative reweighted total variation (wTV) algorithm reduces small streaks well, it is rather inept at eliminating large ones since total variation (TV) regularization is scale-dependent and may regard these streaks as homogeneous

areas. Hence, the main purpose of this paper is to reduce streak artifacts at various scales. We propose the scale-space anisotropic total variation (ssaTV) algorithm, which is derived from wTV, in two different implementations. The first implementation (ssaTV-1) utilizes an anisotropic gradient-like operator which uses 2s neighboring pixels along the streaks’ normal direction at each scale s. The second implementation (ssaTV-2) makes use of anisotropic down-sampling and up-sampling operations, similarly oriented along the streaks’ normal direction, to apply TV regularization at various scales. Experiments on numerical and clinical data demonstrate that both ssaTV algorithms reduce streak artifacts more effectively and efficiently than wTV, particularly when using multiple scales.

Authors with CRIS profile

Involved external institutions

How to cite


Huang, Y., Taubmann, O., Huang, X., Haase, V., Lauritsch, G., & Maier, A. (2018). Scale-Space Anisotropic Total Variation for Limited Angle Tomography. IEEE Transactions on Radiation and Plasma Medical Sciences, 2(4), 307-314.


Huang, Yixing, et al. "Scale-Space Anisotropic Total Variation for Limited Angle Tomography." IEEE Transactions on Radiation and Plasma Medical Sciences 2.4 (2018): 307-314.

BibTeX: Download