Options
2009
Conference Paper
Title
Applications of anisotropic image filters for computing 2D and 3D-fiber orientations
Abstract
Fiber-reinforced materials such as glass and carbon fiber-reinforced polymers or ultra high performance concretes find an increasing number of applications e.g. in construction of bridges, automotive and aerospace engineering. Due to their matrix, such materials typically posses high compressive strengths while the fibers contribute tensile strength to the overall material properties. Measuring these fibers' orientation distribution in images obtained e.g. by micro-computed tomography (mu CT) or scanning acoustic microscopy (SAM) enables an assessment of mechanical properties of a specimen. Unfortunately, image quality is often low, which complicates the segmentation of fibers in such images. We have recently proposed a method for computing fiber orientations directly from gray valued images. This method applies anisotropic Gaussian convolution filters to find the likely local orientation in each pixel. An efficient implementation of this filter operation in 2D and 3D is available. By accumulating the local orientations of foreground pixels in the second order orientation tensor, mean fiber orientations and information on the shape of the fiber orientation distribution can be computed. We here propose a novel sampling scheme for this method, evaluate its accuracy on simulated data and apply it to compute fiber orientations in a mu CT-reconstruction of a carbon fiber-reinforced polymer.