Using the Graph-Cut Method to Segment the Mineralization Area in the Gejiu Region of Yunnan Province, China

Abstract

Geological sampling data often need to be regionally segmented to mark blocks with different geological properties in mathematical geosciences. This paper uses the directed graph model of graph theory to model sampling data and calculate the segmentation of the directed graph by the improved maximum flow minimum cut (max-flow min-cut) algorithm. The proposed energy expressions reflect the continuity of geological properties and the smoothness of segmentation boundaries. A major advantage of this method is that it allows geological experts to share prior knowledge regarding the partitioning of a mineralization area as the seeds of the directed graph. Incorporating this knowledge in the segmentation result via a very intuitive and simple outline operation greatly improves the interactivity and editability of the algorithm. For the integrity of the process framework, this paper also proposes incorporating the characteristic scale of the Hilbert–Huang transformation (HHT) framework as the characteristic parameter to calculate the initial seed, which can achieve automatic regional segmentation.