2009
DOI: 10.1007/978-3-642-10331-5_102
|View full text |Cite
|
Sign up to set email alerts
|

Approximated Curvature Penalty in Non-rigid Registration Using Pairwise MRFs

Abstract: Abstract. Labeling of discrete Markov Random Fields (MRFs) has become an attractive approach for solving the problem of non-rigid image registration. Here, regularization plays an important role in order to obtain smooth deformations for the inherent ill-posed problem. Smoothness is achieved by penalizing the derivatives of the displacement field. However, efficient optimization strategies (based on iterative graph-cuts) are only available for first-order MRFs which contain cliques of size up to two. Higher-or… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

1
9
0

Year Published

2010
2010
2017
2017

Publication Types

Select...
6
3

Relationship

1
8

Authors

Journals

citations
Cited by 16 publications
(10 citation statements)
references
References 19 publications
1
9
0
Order By: Relevance
“…(d) shows the deformed image from Image 1 using the pro- In this example, we compare our proposed QCHR model with other existing registration algorithms using Examples 5-8. We compare our method with TPS [1], LD-DMM [13], diffeomorphic Demons (DDemons) [55], DROP hybrid registration (TPS landmarkmatching + MRF intensity-matching registration) [7,8,6], and Elastix [56,57]. Two measurements are introduced to quantitatively compare the registration results.…”
Section: Hybrid Registration Modelsupporting
confidence: 44%
“…(d) shows the deformed image from Image 1 using the pro- In this example, we compare our proposed QCHR model with other existing registration algorithms using Examples 5-8. We compare our method with TPS [1], LD-DMM [13], diffeomorphic Demons (DDemons) [55], DROP hybrid registration (TPS landmarkmatching + MRF intensity-matching registration) [7,8,6], and Elastix [56,57]. Two measurements are introduced to quantitatively compare the registration results.…”
Section: Hybrid Registration Modelsupporting
confidence: 44%
“…One other limitation of the discrete optimisation approach is that including higher order regularization functions becomes very difficult (Glocker et al, 2009; Kwon et al, 2011), even though the similarity measure can be chosen very flexibly. However, we have found that the advantages of the discrete optimisation framework outweigh the disadvantages for our application.…”
Section: Introductionmentioning
confidence: 99%
“…We selected NMI as the similarity measure for it, which performed better than other metrics in the following tests. In the Drop implementation, FastPD [40] was used as the optimizer, and we selected approximated curvature penalty [42] as the regularization function, which penalizes the approximated second-order derivatives of the displacement field. Moreover, we have empirically found that 500 iterations were sufficient to achieve a registration result with no further improvements and the suitable value for the registration parameter was 10.…”
Section: Parameter Settingsmentioning
confidence: 47%