TY - THES

T1 - Automatic Segmentation Techniques of Profile Measurement for Circle-Line Splining

AU - Hinterleitner, Christoph-Peter

N1 - embargoed until null

PY - 2010

Y1 - 2010

N2 - This thesis addresses techniques for automatic registration and segmentation of 2D geometry measurement data associated with profiled steel tubing. The CAD model consists of a concatenation of many circular arcs and straight line segments, which form the closed profile. The registration and segmentation of the measurement data is a prerequisite to performing a least square fitting of the coupled geometric objects. Prior to fitting the segments of data are associated with specific geometric objects. The cross-section of the geometric objects is measured by a set of light-sectioning sensing heads arranged around the profile. In order to derive quantitative information from the measured data, the data points have to be registered and segmented prior to fitting. In the present work, techniques addressing the following tasks are presented: (1) template matching; (2) data segment assignment; (3) data sorting and (4) determination of transition points between the particular segments. Thereby, CAD data representing the ideal geometry is incorporated. The above process yield segmented and sorted data, ready for least square fitting. All methods are tested and verified with real measurement data, recorded from a automatic measuring system using five light-sectioning heads to capture the cross-section.

AB - This thesis addresses techniques for automatic registration and segmentation of 2D geometry measurement data associated with profiled steel tubing. The CAD model consists of a concatenation of many circular arcs and straight line segments, which form the closed profile. The registration and segmentation of the measurement data is a prerequisite to performing a least square fitting of the coupled geometric objects. Prior to fitting the segments of data are associated with specific geometric objects. The cross-section of the geometric objects is measured by a set of light-sectioning sensing heads arranged around the profile. In order to derive quantitative information from the measured data, the data points have to be registered and segmented prior to fitting. In the present work, techniques addressing the following tasks are presented: (1) template matching; (2) data segment assignment; (3) data sorting and (4) determination of transition points between the particular segments. Thereby, CAD data representing the ideal geometry is incorporated. The above process yield segmented and sorted data, ready for least square fitting. All methods are tested and verified with real measurement data, recorded from a automatic measuring system using five light-sectioning heads to capture the cross-section.

KW - fitting line circle least squares matrix computation

KW - Segmentierung Messdaten Matrizenrechnung Methode der kleinsten Quadrate Kreis Linie

M3 - Diploma Thesis

ER -