Resolving Ambiguities in Reconstructed Grain Maps using Discrete Tomography

The so-called 3DXRD microscope, implemented at the European Synchrotron Radiation Facility in Grenoble, France, utilizes the principle of X-ray diffraction for mapping the crystalline grains within hard materials such as metals or ceramics. Present algorithms, using continuous models, roughly reconstruct the image from diffraction data, but they are often unable to assign unambiguous values to all pixels. We present an approach that resolves these ambiguous pixels by using a Monte Carlo technique that exploits the discrete nature of the problem and utilizes proven methods of discrete tomography. Based on simulations we show that most ambiguities can be successfully resolved.