Heidelberg University

Visualization and
Numerical Geometry Group

ILATO - Improving Limited Angle computed Tomography by Optical data integration

D-A-CH, a joint project financed by the DFG, German Research Foundation, and the SNSF, Swiss National Science Foundation, Switzerland

Members at IWR

Andreas Beyer, Hubert Mara, Susanne Krömker, Hans Georg Bock

Project partners

Yu Liu (ETH Zurich, and Empa), Konrad Wegener (ETH Zurich), Philipp Schütz (Luzern University of Applied Sciences and Arts), Alexander Flisch (Empa), Urs Sennhauser (Empa)
Logo of Empa    
Logo of ETH Zurich
Überlandstrasse 129, 8600 Dübendorf, CH     Rämistrasse 101, 8092 Zurich, CH

Publications

A. Beyer, H. Mara, S. Krömker: ILATO Project: Fusion of Optical Surface Models and Volumetric CT Data, in: Computing Research Repository CoRR, 2014
Abstract, Fulltext PDF

Y. Liu, P. Schütz, A. Flisch, U. Sennhauser: Exploring the limits of limited-angle computed tomography complemented with surface data, in: 11th European Conference on Non-destructive Testing (ECNDT), 2014
Abstract, Fulltext PDF

AluSample with MSII
Multiscale Integral Invariant (MSII) visualization
of a car cylinder showing cuvature (H. Mara)

Short description of the project

ILATO is an acronym of Improving Limited Angle computed Tomography by Optical data integration. By measuring the intensity of X-ray beam after it travels through an object, the attenuation values of a specimen are acquired.

Limited angle sinogram
Schematic view of CT setup and resulting sinogram (from full circle) truncated to represent limited angle scan results (center region without red areas) (Y. Liu)
The 3D information is reconstructed from a series of 2D projections mostly on a circular trajectory. This is accomplished via inverse radon transformation. Limited angle computed tomography is indicated whenever measurements from certain angles are impossible. Incomplete data from the limited angle CT measurement poses problems to the reconstruction which introduces blurred edges and other artifacts.







Trajectory around a sample
Circular trajectory from CT scan with pyramids representing X-ray source positions
Surface representations acquired from optical scans are processed according to the corresponding CT trajectory to inspect all rays intersecting the object and to calculate the exact ray length in material. This information enables enhanced reconstruction to deliver a superior volumetric representation of the specimen.







Sample
Surface extraction from CT data with VESTA vs commercial marching cube
A surface mesh acquisition of the object obtained via structured light optical scan is not limited in angle but cannot reveal the volumetric structure inside. To support the volumetric data reconstruction algorithm a surface mesh acts as a mask defining boundaries of the reconstructed image. For registration purposes of the outer boundary of CT data with the surface of the optical scan a Volume-Enclosing Surface ex-Traction Algorithm (VESTA; B. R. Schlei) is implemented and tested against a commercial marching cube algorithm.



Limited Angle CT vs Optical Scan
Left: Typical limited angle CT artifacts due to incomplete information for reconstruction
Right: Optical scan with correct dimensions
The goal of the project is to correct certain defects in reconstructing limited angle CT scans which occur from insufficient information. As attenuation measurement based imaging techniques collect data only as footprints of each ray and limited angle CT does not provide a full set of projection directions, reconstruction algorithms tend to introduce an elongation which can be observed as an anisotropic magnification effect in the resulting representation of the object.


The project focuses on the registration of optically acquired surfaces with data acquired from computed tomography in order to enhance image quality and shorten acquisition times in X-ray based industrial quality inspection.


Sample with deviation
Orthogonal Projection:
Geometrical deformations of specimen introduced by artifacts in CT workflow (A. Beyer)
The challenges are the correct setup,
the registration of data sets created with considerably different imaging techniques, and the improvement of the applied methods.

All the extracted information is used to gain an improved visualization in the sense of less noise and sharper edges especially in the areas where the limited angle reconstruction suffers from insufficient information.




Please note: We explicitly point out that all copyrights and rights of use of the material and data provided on this website are property of IWR - Heidelberg University. Any commercial utilisation of our data is strictly prohibited. Furthermore, the dissemination of any data or images as well as physical copies or material produced on the basis of the original data, requires our written consent on a case by case basis. Without our agreement, it is also expressly forbidden to disseminate or publish any kind of modifications of the data or images provided. Downloading any of the material made available on this website implies your full acceptance of our terms and conditions as described above.


S. Krömker, 
Last Update: 07.11.2014
zum Seitenanfang