Volume 28, Issue 1 (January 2000)
Efficient Mitigation of Edge Effects in Nearest-Neighbor Analysis
The properties of soils and other civil engineering construction materials are influenced by the microstructural arrangement of the particles. The characterization of the spatial arrangement and distribution of microstructural features are therefore of importance if the mechanical properties of materials commonly used in civil engineering are to be better understood. Numerous statistical methods have been developed to characterize the spatial arrangement of microstructural features. However, care must be taken in implementing these methods using automated imaging systems to analyze finite quantities of features so that bias due to edge effects and other factors is not introduced. A new method for mitigating edge effects when nearest-neighbor analysis is being performed is proposed in this paper. The method is validated using synthetic square and triangular patterns for which the spatial arrangement free of edge effects (unbiased) is already known. Use of the method to analyze a real dataset representing the peripheral packing arrangement of sand particles in triaxial specimens is described. The results indicate that the new method is able to mitigate edge effects and offer substantial savings in terms of efficiency of data collection and analysis.