1 resultado para Image Classification
em Digital Commons - Michigan Tech
Resumo:
Quantifying belowground dynamics is critical to our understanding of plant and ecosystem function and belowground carbon cycling, yet currently available tools for complex belowground image analyses are insufficient. We introduce novel techniques combining digital image processing tools and geographic information systems (GIS) analysis to permit semi-automated analysis of complex root and soil dynamics. We illustrate methodologies with imagery from microcosms, minirhizotrons, and a rhizotron, in upland and peatland soils. We provide guidelines for correct image capture, a method that automatically stitches together numerous minirhizotron images into one seamless image, and image analysis using image segmentation and classification in SPRING or change analysis in ArcMap. These methods facilitate spatial and temporal root and soil interaction studies, providing a framework to expand a more comprehensive understanding of belowground dynamics.