Characterization of Multispectral Visible and X-ray CT Imagery of Hydromorphic Soils

In this project, semivariance analysis was applied on multispectral soil thin section imagery and X-ray CT imagery to delineate pedogenic variability of hydromorphic soils and for guiding object-based segmentation. Semivariogram indices were computed from multispectral visible channels to maximize the information obtained from optical light imaging of soil thin section images and multi-energy X-ray CT images. The use of dual angle cross polarization, in multispectral soil thin section imagery, is capable of extracting heterogeneous information. Object-based image segmentation showed satisfying result guided by semivariance analysis, under various scenarios regarding Gleysolic classification and horizonation. Particular mineral separation from soil matrix in CT greyscale images is challengeable according to soil composition complexity and the closeness of Hounsfield units of mineral oxidation. Based on relative attenuation in histograms of CT images obtained by multi-energy X-ray spectra, the suggestion of using multi-energy X-ray CT imagery is given to simplify material-specific segmentation.