Univ.-Prof. Dipl.-Ing. Dr. Bernhard Zagar
|May 12th, 2014|
In steel industry the quality standard of production, storage and shipping of steel coils is very high. An important quality index of steel coils is their roundness.
At the beginning of this thesis various indices to describe the roundness of the coil are presented and a simulation of the ellipticity of a coil is discussed.
For the image acquisition a software was developed in C# to control the AVT Firewire camera.
Using digital image processing data (bended curve segments; see figure 1) is extracted, which is used to determine the center and ovality (see figure 2) of the coil with the specially for this problem designed concentric ellipse fit estimator.
To validate the results of this estimator the Kanatani Cramer-Rao lower bound and the Monte Carlo method were used, which showed that the estimator leads to very good results when proper input data is available. A detailed error analysis shows the influences of the input data and image acquisition setup, which affects the estimation results.
It shows that the computer based analysis can be of great assistance to the human observer and leads to reliable results.
Keywords: coil, roundness, ovality, ellipse, estimation, quality