Multiphasic Quantification of Soil Morphology using High-Resolution X-ray Computed Tomography

X射线计算机层析成像(CT)已成为了解土壤组分三维形态的重要工具。这项技术提供了对土壤物理性质的独特见解，特别是其水动力学行为和管理做法对结构质量的影响。土壤的典型CT图像的主要局限性之一是其单色性，限制了我们区分各种固体和有机固体的能力。部分体积的问题进一步增加了分割和分类的复杂性，由此由图像的给定单位(体素)表示的值可能由一种以上的材料产生(称为部分体积效应)。此外，我们处理3D图像的能力落后于2D图像(这主要是由遥感应用推动的)；部分原因也是因为对3D图像的计算要求更高。因此，将X射线CT应用于土壤的有效标准化方法尚未实现。**当前的研究计划将专注于提高我们区分主要矿物和有机固相的能力，这些固相通常出现在土壤中，或可能作为污染物存在。将特别注意通过扫描能量和金属预过滤的适当组合来改进X射线束的成分，以便在扫描阶段最大限度地区分不同的阶段。此外，还将采用先进的基于对象的图像分离技术，使其适用于三维X射线CT图像。将特别注意优化水成土特征的量化，特别是铁和锰的分离。最后，将继续改进分类图像的形态分析方法和原始灰度级图像的高级空间分析方法，包括多重分形和熵技术，以利用改进的扫描和分割技术。这项研究将主要在圭尔夫大学的土壤成像实验室进行，但也将利用X射线CT成像和图像处理方面的国际专业知识网络，并提供研究生和本科生级别的培训机会。

X-ray computed tomography (CT) has become an important tool for understanding the three-dimensional (3D) configuration of soil components. This technology has provided unique insight into the physical nature of soil, especially its hydrodynamic behaviour and the impact of management practices on structural quality. One of the principle limitations of typical CT imagery of soil is its monochromatic nature, limiting our ability to distinguish among the various solid and organic solids. The issue of partial volumes, whereby the value expressed by a given unit of the imagery (voxel) may result from more than one material (referred to as the partial volume effect) further introduces complexity to segmentation and classification. Moreover, our capacity to process 3D imagery lags behind that for 2D imagery (which has been driven primarily by remote sensing applications); in part, this is also due to greater computational requirements for 3D imagery. As such, effective standardization of methodologies for the application of X-ray CT to soil has yet to be achieved.**The current research program will focus on contributing our ability to discriminate the major mineral and organic solid phases that commonly occur naturally in soils, or may be present as contaminants. Specific attention will be given to refining the composition of the X-ray beam, through appropriate combinations of scanning energy and metal pre-filtering, to maximize the discrimination among phases during the scanning stage. This will be complemented by the adaptation of advanced object-based image segregation techniques to 3D X-ray CT imagery. Specific attention will be given to optimizing the quantification of hydromorphic pedofeatures, specifically Fe and Mn segregations. Finally, methods for morphometric analysis of classified imagery and advanced spatial analysis, including multi-fractal and entropy techniques, of the original grey-scale imagery will continue to be refined to take advantage of improved scanning and segmentation techniques. This research will be conducted primarily in the Soil Imaging Laboratory of the University of Guelph, but will also engage an international network of expertise on X-ray CT imaging and image processing, and provide both post-graduate and undergraduate level training opportunities.