Prakt. Met. Sonderband 38 (2006) 127 
ndet. Mit AUTOMATIC REGISTRATION OF CORRESPONDING 
Sbildeten FRACTURE SURFACES 
R. Danzl*, S. Scherer*, O. Kolednik** 
Alicona Imaging GmbH, 8074 Grambach, Austria 
* Erich Schmid Institute of Materials Science, Austrian Academy of Sciences, 8700 
Leoben, Austria 
ABSTRACT 
A novel algorithm has been developed for the automatic registration of fracture surfaces 
which is included within a framework for the calculation of local fracture surface 
parameters. First 3D models are reconstructed of corresponding regions on both halves of 
a fractured specimen using stereoscopic images from the scanning electron microscope. 
The 3D models are then automatically registered to each other so that regions with little 
plastic deformations (fatigue fracture area) fit perfectly to each other. Afterwards surface 
profiles can be drawn in one of the input images and the corresponding surface profile of 
the second half is automatically determined. These profiles are used to calculate local 
fracture toughness parameters such as the critical crack tip opening displacement COD.. 
alytischer To check the registration procedure, it is applied on two samples with different fracture 
So wurde behavior. The automatic method is compared to results obtained from traditional manual 
ilung die search of corresponding surface profiles, demonstrating the accuracy and benefit of the 
hode war automatic method. 
essdaten 
en ist die 
ant. 1. INTRODUCTION 
1andenen 
bewertet For quite some time stereo photogrammetry has been used in the field of fracture research 
/erteilung to quantitatively analyse fracture surfaces and to derive characteristic parameters for 
größen in different materials [4,5,8]. One of the most important is the critical crack tip opening 
gen und displacement (COD) [8,13], a local measure for the fracture toughness of the material. In 
contrast to global measures, local parameters such as COD; have the advantage that they 
better describe the fracture properties of inhomogeneous materials and give hints on the 
influence of microstructure on the local fracture toughness [6,7]. 
So far the determination of COD; has required a lot of user input: Stereoscopic images 
taken from the fracture surface in the scanning electron microscope (SEM) are analyzed 
einbaren using a digital 3D reconstruction software [9,13]. As a result a 3D model of the imaged 
area is obtained. Afterwards the opposite 3D model is analyzed from the corresponding 
fracture surface region on the second half of the broken specimen. In order to determine 
3 COD, exactly corresponding profiles are manually drawn perpendicular to the pre-crack 
front in the images of the two corresponding fracture surfaces and vertically aligned [8,13], 
see Fig. 1. Depending on the character of the fracture surfaces, this search for 
f corresponding profiles may require a lot of time and can be a big source of errors. 
In this work we propose a method which performs the search for corresponding surface 
profiles (semi-)automatically and thus drastically alleviates this task. The basic idea is to 
register the 3D models of the two corresponding fracture surfaces so that the misfit 
; between them reaches a minimum. In order to deal with regions of bad correspondence. 
OD