Abstract:
The Fry Method, a powerful technique for strain estimation, requires precise identification of the ellipse in the central void of the Fry plot. Several geometric, algebraic and numerical methods exist for identification of the central vacancy best-fit ellipse. Most of these methods are semi-automatic and needs some level of manual interpretation. The manual interpretation also increases the degree of biasing and subjectivity. Some of the methods for estimating strain use user-defined constants (Waldron and Wallace, 2007). We have introduced a new image analysis method for estimating the strain by Fry method.
We use the Gaussian blur which is analogous to viewing the Fry plot through a translucent screen so that the high density region can be distinguished in the form of an elliptical ring, from the other region thus enabling the accurate fitting of an ellipse. The method uses 'Threshold' for extracting the central vacancy and then fits an ellipse using the Region of Interest selection algorithm. We have demonstrated the validity and efficacy of our method on computer simulated and natural examples. Furthermore we have also demonstrated the effect of sorting, point size and the variation in Gaussian blur on the results of the Image-analysis method. We have also compared our method with the results obtained by the other methods (Point-count density method and the continuous function method) on distorted samples with different degrees of sorting. We have also done an extensive error analysis of the image-analysis, point-count density, continuous function and retrodeformation methods for comparison.
