SELECTIVE SEGMENTATION OF SHALLOW ELECTRICAL PROFILING DATA
A. G. Zlobina, I. V. Zhurbin, O. M. Nemtsova
Physical Engineering Institute of the Ural Branch of the Russian Academy of Sciences, 426000, Izhevsk, ul. Kirova, 132
Keywords: электропрофилирование, компьютерное моделирование, нечёткая кластеризация, векторные картины преимущественного направления, метод селективной сегментации, граница объекта, диапазон глубины залегания, electrical profiling, computer simulation, fuzzy clustering, vector pictures of the primary direction, selective segmentation method, object boundary, range of the depth of occurrence
Abstract
A selective segmentation method is developed that correctly recovers the boundaries of an object in the plan and estimates the depth range of its location. This method assumes the processing of electrical profiling data: shallow electrical profiling and vector patterns of dominant directions of changes in the medium resistance. The boundary of the object is estimated on the basis of using a fuzzy c-means clustering algorithm, adapted to analyze the multidimensional electrical profiling data. The depth range of the object location in the soil is estimated on the basis of using the scalar product function with respect to the vector patterns of dominant directions, adjacent in depth. The effectiveness of the method is confirmed by the computer simulation results and the example of processing of the full-scale experiment on an archaeological site, which is a medieval fortified settlement.
|
