Detection and Size Identification of Individual Grains from Geological Images
Abstract
In this thesis, we focus on identifying the grain size distribution from geological images of model geometries. Since grain size can be interpreted as a scale or frequency, we employ the Fourier transform and continuous wavelet transform. Based on observations of synthetic grain images, we design tailored algorithms to detect the statistics of individual grains. Subsequently, we modify these algorithms and analyze experimental images. The detected grain size distribution histograms are then compared with those obtained from other methods to evaluate the performance of our algorithm.
Our experiments demonstrate that both methods can extract grain size information from images. However, the continuous wavelet transform outperforms the Fourier transform as it captures both scale and location information. Potential applications of our algorithm include geometric segmentation and pore identification in images of porous media.