Structural Gaussian priors for Bayesian CT reconstruction of subsea pipes

Christensen, Silja Westphal; Riis, Nicolai A. B.; Uribe, Federico; Jørgensen, Jakob Sauer

doi:10.1080/27690911.2023.2224918

Cited by 1 publication

(7 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We assume some structural information about the pipe is known a-priori from design specifications or previous reference scans. Alternatively, the structural information can be obtained via a deterministic reconstruction as described in [28]. Specifically, we assume the pipe consists of layers with approximately known layer boundaries, and that the materials in these layers are well known, such that we can obtain good a-priori estimates of their linear attenuation coefficients.…”

Section: Bayesian Formulationmentioning

confidence: 99%

“…Specifically, we assume the pipe consists of layers with approximately known layer boundaries, and that the materials in these layers are well known, such that we can obtain good a-priori estimates of their linear attenuation coefficients. The structural Gaussian prior (SGP) was proposed in [28] and it exploits exactly this information. Therefore, it is employed here as the pipe prior.…”

Section: Bayesian Formulationmentioning

confidence: 99%

“…Each square-root precision matrix 5 has the form R k = √ δ k M k , where δ k is a scalar precision parameter that controls the strength of the prior, and M k contains the structure of the square-root precision matrix. For more details see [28]. To illustrate the strong information encoded in the SGP, we show two realizations from (9) in figure 2, where the prior is specified according to the pipe described in section 4.2.…”

Section: Bayesian Formulationmentioning

confidence: 99%

“…We use linear RTO ('randomize-thenoptimize') [43] to sample efficiently from these high dimensional Gaussian distributions, as described in the following. A realization z * from z | d, b = b obs is obtained by solving the following system of linear equations [28]:…”

Section: Posterior Samplingmentioning

confidence: 99%

“…To promote sparsity and pipe structure in the Bayesian setting we must choose appropriate priors for both images. The general internal structure and materials of the pipes are typically well known and this information can be used to formulate a prior for the pipe reconstruction [28]. The other important piece of information we have, is that we expect few and small defects in the pipe, i.e.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

A Bayesian approach for CT reconstruction with defect detection for subsea pipelines

Christensen,

Riis,

Pereyra

et al. 2023

Inverse Problems

Self Cite

View full text Add to dashboard Cite

Subsea pipelines can be inspected via 2D cross-sectional x-ray computed tomography (CT). Traditional reconstruction methods produce an image of the pipe’s interior that can be post-processed for detection of possible defects. In this paper we propose a novel Bayesian CT reconstruction method with built-in defect detection. We decompose the reconstruction into a sum of two images; one containing the overall pipe structure, and one containing defects, and infer the images simultaneously in a Gibbs scheme. Our method requires that prior information about the two images is very distinct, i.e. the first image should contain the large-scale and layered pipe structure, and the second image should contain small, coherent defects. We demonstrate our methodology with numerical experiments using synthetic and real CT data from scans of subsea pipes in cases with full and limited data. Experiments demonstrate the effectiveness of the proposed method in various data settings, with reconstruction quality comparable to existing techniques, while also providing defect detection with uncertainty quantification.

show abstract

Section: Bayesian Formulationmentioning

confidence: 99%

Section: Bayesian Formulationmentioning

confidence: 99%