Oliver Wirjadi
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Oliver Wirjadi has a total of 62 co-authors. Here are the top co-authors by number of shared publications:
- Katja Schladitz
Fraunhofer-Institut für Techno-und Wirtschaftsmathematik (ITWM) Fraunhofer-Platz 1 67663 Kaiserslautern Germany
7 shared publications - Alexander Rack
ESRF – The European Synchrotron, Grenoble, France
5 shared publications - Claudia Redenbach
RPTU University Kaiserslautern-Landau
4 shared publications Heiko Andrä
Fraunhofer ITWM
3 shared publicationsThomas Breuel
Department of Computer Science, University of Kaiserslautern, 67663 Kaiserslautern, Germany
3 shared publicationsBrigitte Dix
Fraunhofer WKI
2 shared publications
Latest Publications
- Open Access
- 7 December 2019
- Published by Slovenian Society for Stereology and Quantitative Image Analysis in Image Analysis and Stereology
- Vol. 38 (3) , 245-253
Endpoint Detection of Partially Overlapping Straight Fibers using High Positive Gaussian Curvature in 3D images
This paper introduces a method for detecting endpoints of partially overlapping straight fibers in three-dimensional voxel image data. The novel approach directly determines fiber endpoints without the need for more expansive single-fiber segmentation. In the context of fiber-reinforced polymers, endpoint information is of practical significance as it can indicate potential damage in endless fiber systems, or can serve as input for estimating statistical fiber length distribution. We tackle this challenge by exploiting Gaussian curvature of the surface of the fibers. Fiber endpoints have high positive curvature, allowing one to distinguish them from the rest of a structure. Accuracy data of the proposed method are presented for various data sets. For simulated fiber systems with fiber volume fractions of less than 20 %, true positive rates above 94 % and false positive rates below 5 % are observed. Two well-resolved real data sets show a reduction of the first rate to 90.3 % and an increase of the second rate to 13.1 %.
- 7 December 2019
- Published by Slovenian Society for Stereology and Quantitative Image Analysis in Image Analysis and Stereology
- Vol. 38 (3) , 245-253
Open AccessEndpoint Detection of Partially Overlapping Straight Fibers using High Positive Gaussian Curvature in 3D images
This paper introduces a method for detecting endpoints of partially overlapping straight fibers in three-dimensional voxel image data. The novel approach directly determines fiber endpoints without the need for more expansive single-fiber segmentation. In the context of fiber-reinforced polymers, endpoint information is of practical significance as it can indicate potential damage in endless fiber systems, or can serve as input for estimating statistical fiber length distribution. We tackle this challenge by exploiting Gaussian curvature of the surface of the fibers. Fiber endpoints have high positive curvature, allowing one to distinguish them from the rest of a structure. Accuracy data of the proposed method are presented for various data sets. For simulated fiber systems with fiber volume fractions of less than 20 %, true positive rates above 94 % and false positive rates below 5 % are observed. Two well-resolved real data sets show a reduction of the first rate to 90.3 % and an increase of the second rate to 13.1 %.
- 31 July 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Visualization and Computer Graphics
- Vol. 25 (10) , 3032-3041
Empirical Comparison of Curvature Estimators on Volume Images and Triangle Meshes
In the development of graphical algorithms, choosing an appropriate data representation plays a pivotal role. Hence, there is a need for studies that support corresponding decision making. Here, we investigate curvature estimation based on two discrete representations - volume images and triangle meshes - and present a comprehensive cross-comparison. For doing so, four carefully selected geometries, represented as implicit functions, have been discretized to volume images and triangle meshes in different resolutions on a comparable scale. Afterwards, implementations available in open-source libraries (CGAL, DIPimage, libigl, trimesh2, VTK) and our own implementation of a relevant paper were applied to them and the resulting estimations of mean and Gaussian curvature were compared in terms of quality and runtime. Independent of the underlying discrete representation, all estimators generated similar errors, but overall, mesh-based methods allowed for more accurate estimations. We measured a maximum normalized mean absolute error difference of 6.36% between the most precise mesh-based method compared to corresponding image-based methods when considering only discretizations of sufficient resolution. In terms of runtime, methods working on triangle meshes were faster when geometries had a small surface density. For geometries with larger surface densities, which is fairly common when considering real data, e.g. in material or medical science, the runtimes for both representations were similar.
- 31 July 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Visualization and Computer Graphics
- Vol. 25 (10) , 3032-3041
Empirical Comparison of Curvature Estimators on Volume Images and Triangle Meshes
In the development of graphical algorithms, choosing an appropriate data representation plays a pivotal role. Hence, there is a need for studies that support corresponding decision making. Here, we investigate curvature estimation based on two discrete representations - volume images and triangle meshes - and present a comprehensive cross-comparison. For doing so, four carefully selected geometries, represented as implicit functions, have been discretized to volume images and triangle meshes in different resolutions on a comparable scale. Afterwards, implementations available in open-source libraries (CGAL, DIPimage, libigl, trimesh2, VTK) and our own implementation of a relevant paper were applied to them and the resulting estimations of mean and Gaussian curvature were compared in terms of quality and runtime. Independent of the underlying discrete representation, all estimators generated similar errors, but overall, mesh-based methods allowed for more accurate estimations. We measured a maximum normalized mean absolute error difference of 6.36% between the most precise mesh-based method compared to corresponding image-based methods when considering only discretizations of sufficient resolution. In terms of runtime, methods working on triangle meshes were faster when geometries had a small surface density. For geometries with larger surface densities, which is fairly common when considering real data, e.g. in material or medical science, the runtimes for both representations were similar.
- 20 September 2016
- Published by Wiley in Journal of Microscopy
- Vol. 265 (2) , 159-168
Bayesian model for detection and classification of meningioma nuclei in microscopic images
Image segmentation aims to determine structures of interest inside a digital picture in biomedical sciences. State-of-the art automatic methods however still fail to provide the segmentation quality achievable by humans who employ expert knowledge and use software to mark target structures on an image. Manual segmentation is time-consuming, tedious and suffers from interoperator variability, thus not serving the requirements of daily use well. Therefore, the approach presented here abandons the goal of full-fledged segmentation and settles for the localization of circular objects in photographs (10 training images and 20 testing images with several hundreds of nuclei each). A fully trainable softcore interaction point process model was hence fit to the most likely locations of nuclei of meningioma cells. The Broad Bioimage Benchmark Collection/SIMCEP data set of virtual cells served as controls. A ‘colour deconvolution’ algorithm was integrated to determine (based on anti-Ki67 immunohistochemistry) which real cells might have the potential to proliferate. In addition, a density parameter of the underlying Bayesian model was estimated. Immunohistochemistry results were ‘simulated'for the virtual cells. The system yielded true positive (TP) rates in the detection and classification of real nuclei and their virtual counterparts. These hits outnumbered those obtained from the public domain image processing software ImageJ by 10%. The method introduced here can be trained to function not only in medicine and morphology-based systems biology but in other application domains as well. The algorithm lends itself to an automated approach that constitutes a valuable tool which is easy to use and generates acceptable results quickly. Microscopes are used in biomedical sciences e.g. in the context of pathology. There, one task is for the pathologist to judge the severity of a resected tumor. Today, digital pictures from such microscopes are readily available and (in theory) open the possibility for automation by means of analyzing these data algorithmically. Yet, even though manual analysis of such images by human experts is time-consuming, tedious, and suffers from inter-operator variability, automated algorithms nevertheless do not yet reach the standards required for their routine clinical usage. One reason for this is the difficutly of a process called “segmentation”, which is one of the algorithmic steps comprising a potential automated, algorithmic approach. The term “segmentation” describes the identification of different objects in an image (e.g., tumor cells separated from other visible tissue in a pathological micrograph). There, the classical algorithmic approach is to identify “regions” in digital images which show objects (e.g., tumor cells). Even though a large variety of theoretically well-founded approaches to this problem exist in the relevant literature, no solution seems to be sufficiently reliable, to date. In this work, we describe how in certain situations such as the computer-aided grading of meningioma, the problem can be simplified, thus avoiding the difficult segmentation step. Instead, we describe and further automate a statistical method to find only the loci of tumor cells and (separately) how to employ a method described elsewhere to decide whether cells have potential to proliferate in terms of Ki-67 immunohistochemistry. Based on a collection of micrographs from different mengingioma and basd on a public benchmark database, we are able to demonstrate the usefulness of our methods.
- 20 September 2016
- Published by Wiley in Journal of Microscopy
- Vol. 265 (2) , 159-168
Bayesian model for detection and classification of meningioma nuclei in microscopic images
Image segmentation aims to determine structures of interest inside a digital picture in biomedical sciences. State-of-the art automatic methods however still fail to provide the segmentation quality achievable by humans who employ expert knowledge and use software to mark target structures on an image. Manual segmentation is time-consuming, tedious and suffers from interoperator variability, thus not serving the requirements of daily use well. Therefore, the approach presented here abandons the goal of full-fledged segmentation and settles for the localization of circular objects in photographs (10 training images and 20 testing images with several hundreds of nuclei each). A fully trainable softcore interaction point process model was hence fit to the most likely locations of nuclei of meningioma cells. The Broad Bioimage Benchmark Collection/SIMCEP data set of virtual cells served as controls. A ‘colour deconvolution’ algorithm was integrated to determine (based on anti-Ki67 immunohistochemistry) which real cells might have the potential to proliferate. In addition, a density parameter of the underlying Bayesian model was estimated. Immunohistochemistry results were ‘simulated'for the virtual cells. The system yielded true positive (TP) rates in the detection and classification of real nuclei and their virtual counterparts. These hits outnumbered those obtained from the public domain image processing software ImageJ by 10%. The method introduced here can be trained to function not only in medicine and morphology-based systems biology but in other application domains as well. The algorithm lends itself to an automated approach that constitutes a valuable tool which is easy to use and generates acceptable results quickly. Microscopes are used in biomedical sciences e.g. in the context of pathology. There, one task is for the pathologist to judge the severity of a resected tumor. Today, digital pictures from such microscopes are readily available and (in theory) open the possibility for automation by means of analyzing these data algorithmically. Yet, even though manual analysis of such images by human experts is time-consuming, tedious, and suffers from inter-operator variability, automated algorithms nevertheless do not yet reach the standards required for their routine clinical usage. One reason for this is the difficutly of a process called “segmentation”, which is one of the algorithmic steps comprising a potential automated, algorithmic approach. The term “segmentation” describes the identification of different objects in an image (e.g., tumor cells separated from other visible tissue in a pathological micrograph). There, the classical algorithmic approach is to identify “regions” in digital images which show objects (e.g., tumor cells). Even though a large variety of theoretically well-founded approaches to this problem exist in the relevant literature, no solution seems to be sufficiently reliable, to date. In this work, we describe how in certain situations such as the computer-aided grading of meningioma, the problem can be simplified, thus avoiding the difficult segmentation step. Instead, we describe and further automate a statistical method to find only the loci of tumor cells and (separately) how to employ a method described elsewhere to decide whether cells have potential to proliferate in terms of Ki-67 immunohistochemistry. Based on a collection of micrographs from different mengingioma and basd on a public benchmark database, we are able to demonstrate the usefulness of our methods.
- 29 December 2015
- Published by Springer Nature in Materials and Structures
- Vol. 49 (10) , 4003-4012
Estimation of fiber orientation and fiber bundles of MDF
- 29 December 2015
- Published by Springer Nature in Materials and Structures
- Vol. 49 (10) , 4003-4012
Estimation of fiber orientation and fiber bundles of MDF
- 1 November 2015
- Published by Elsevier in Graphical Models
- Vol. 82, 123-136
Gaussian curvature using fundamental forms for binary voxel data
- 1 November 2015
- Published by Elsevier in Graphical Models
- Vol. 82, 123-136
Gaussian curvature using fundamental forms for binary voxel data
- 1 December 2014
- Published by Elsevier in Mechanics of Materials
- Vol. 79, 73-84
Numerical prediction of the stiffness and strength of medium density fiberboards
- 1 December 2014
- Published by Elsevier in Mechanics of Materials
- Vol. 79, 73-84
Numerical prediction of the stiffness and strength of medium density fiberboards
- 25 November 2014
- Published by Wiley in GAMM-Mitteilungen
- Vol. 37 (2) , 281-305
Image analysis for microstructures based on stochastic models
The development of modern high‐performance materials requires a deeper understanding of the complex relations between a material's microstructure geometry and its macroscopic properties. Analysis of three‐dimensional image data combined with stochastic microstructure modelling is a promising approach to study these relations. Motivated by two typical application examples, a fibre reinforced polymer and a closed polymer foam, we introduce versatile model classes from stochastic geometry. We explain how their basic geometric characteristics can be estimated from tomographic image data. Finally, linking the estimated values to the model parameters, stochastic models are fitted to the observed microstructures.
- 25 November 2014
- Published by Wiley in GAMM-Mitteilungen
- Vol. 37 (2) , 281-305
Image analysis for microstructures based on stochastic models
The development of modern high‐performance materials requires a deeper understanding of the complex relations between a material's microstructure geometry and its macroscopic properties. Analysis of three‐dimensional image data combined with stochastic microstructure modelling is a promising approach to study these relations. Motivated by two typical application examples, a fibre reinforced polymer and a closed polymer foam, we introduce versatile model classes from stochastic geometry. We explain how their basic geometric characteristics can be estimated from tomographic image data. Finally, linking the estimated values to the model parameters, stochastic models are fitted to the observed microstructures.
- 14 July 2014
- Published by Wiley
- p. 35-40
Geometric and Mechanical Modeling of Fiber‐Reinforced Composites
- 14 July 2014
- Published by Wiley
- p. 35-40
Geometric and Mechanical Modeling of Fiber‐Reinforced Composites
- 14 July 2014
- Published by Walter de Gruyter GmbH in International Journal of Materials Research
- Vol. 105 (7) , 645-654
Characterization of multilayer structures in fiber reinforced polymer employing synchrotron and laboratory X-ray CT
Specimens of carbon or glass fiber reinforced polymer can be imaged using both conventional laboratory X-ray micro-computed tomography equipment and synchrotron light sources. The image quality when using intense (partially) coherent synchrotron light is still superior, especially when applying phase-retrieval algorithms. In the resulting volume images, the fiber direction distribution and other mechanically relevant parameters such as volume fractions or layer thickness can be determined. In this contribution, we will demonstrate how fiber direction results can be used to detect regions with locally different fiber orientations in carbon or glass fiber reinforced polymer which arise in the molding process of such samples. To this end, we evaluate the three-dimensional fiber orientation tensor locally across the thickness of different specimens. For each resulting individual layer, we can automatically detect the layer thickness and the preferred fiber direction. These methods have been successfully applied to various commercial specimens. We will demonstrate results on volume images of samples from both synchrotron and laboratory micro-computed tomography and discuss the specific advantages and disadvantages in this application.
- 14 July 2014
- Published by Walter de Gruyter GmbH in International Journal of Materials Research
- Vol. 105 (7) , 645-654
Characterization of multilayer structures in fiber reinforced polymer employing synchrotron and laboratory X-ray CT
Specimens of carbon or glass fiber reinforced polymer can be imaged using both conventional laboratory X-ray micro-computed tomography equipment and synchrotron light sources. The image quality when using intense (partially) coherent synchrotron light is still superior, especially when applying phase-retrieval algorithms. In the resulting volume images, the fiber direction distribution and other mechanically relevant parameters such as volume fractions or layer thickness can be determined. In this contribution, we will demonstrate how fiber direction results can be used to detect regions with locally different fiber orientations in carbon or glass fiber reinforced polymer which arise in the molding process of such samples. To this end, we evaluate the three-dimensional fiber orientation tensor locally across the thickness of different specimens. For each resulting individual layer, we can automatically detect the layer thickness and the preferred fiber direction. These methods have been successfully applied to various commercial specimens. We will demonstrate results on volume images of samples from both synchrotron and laboratory micro-computed tomography and discuss the specific advantages and disadvantages in this application.
- 14 July 2014
- Published by Walter de Gruyter GmbH in International Journal of Materials Research
- Vol. 105 (7) , 702-708
Microstructural analysis of a C/SiC ceramic based on the segmentation of X-ray phase contrast tomographic data
This paper is dedicated to the analysis of 3D data of carbon fiber reinforced silicon carbide ceramics. In the production process of C/SiC, a porous carbon preform reinforced with bundles of carbon fibers is infiltrated with liquid silicon at approximately 1 500 °C. The reaction between liquid silicon and carbon creates a layer of silicon carbide while the silicon vanishes almost completely. This material is able to withstand extremely high temperatures and it is extremely tough with respect to fracture. To increase the efficiency of the costly and time consuming production process, methods for monitoring the quality of the material are helpful. For instance, the thickness of the silicon carbide layer is a valuable measure. Moreover, due to different coefficients of thermal expansion of the components, typically cracks appear during the production process. For effective analysis 3D high resolution image data are necessary that can be acquired by synchrotron computed tomography. In a first image processing step, we segment the 3D image data with a convex optimization approach incorporating spatial regularity. Further, we work on the detection of cracks using an eigenvalue analysis of the 3D Hessian matrix determined in each pixel.
- 14 July 2014
- Published by Walter de Gruyter GmbH in International Journal of Materials Research
- Vol. 105 (7) , 702-708
Microstructural analysis of a C/SiC ceramic based on the segmentation of X-ray phase contrast tomographic data
This paper is dedicated to the analysis of 3D data of carbon fiber reinforced silicon carbide ceramics. In the production process of C/SiC, a porous carbon preform reinforced with bundles of carbon fibers is infiltrated with liquid silicon at approximately 1 500 °C. The reaction between liquid silicon and carbon creates a layer of silicon carbide while the silicon vanishes almost completely. This material is able to withstand extremely high temperatures and it is extremely tough with respect to fracture. To increase the efficiency of the costly and time consuming production process, methods for monitoring the quality of the material are helpful. For instance, the thickness of the silicon carbide layer is a valuable measure. Moreover, due to different coefficients of thermal expansion of the components, typically cracks appear during the production process. For effective analysis 3D high resolution image data are necessary that can be acquired by synchrotron computed tomography. In a first image processing step, we segment the 3D image data with a convex optimization approach incorporating spatial regularity. Further, we work on the detection of cracks using an eigenvalue analysis of the 3D Hessian matrix determined in each pixel.