Abstract:

In consideration of the importance of multi-scale characterization of coal fractures for understanding the mechanisms of multi-scale fluid transportation, we propose a new approach for a detailed-accurate representation of pore structure of coal rock on CT image with multiple scales based on image processing techniques. The multi-scheme segmentations are employed to extract the object of interest. Representation of the relations among the interested objects, establishment of such image descriptors as the regions, the borders, the statistical moments, and fractal characteristics of the objects are systematically investigated. The proposed approach is applied to characterize the macro- and meso- fractures on the CT image of the coal rock samples. The obtained results show that the object segmentation for macro-fractures can be achieved by thresholding segmentation method. For extraction of the meso-fractures, complex segmentation strategies should be adopted, for example, the Sobel gradient operator with watershed transformation in our study. The multi-scale and detailed-accurate descriptions are achieved on pore structure of the coal rock by using chain code representation, image descriptor and fractal descriptor in Euclidean and fractal spaces.

Key words: coal rock, pore structure, computerized tomography, image description, multi-scale analysis