Extraction of mineralized indicator minerals using ensemble learning model optimized by SSA based on hyperspectral image

Abstract Mineralized indicator minerals are an important geological and mineral exploration indicator. Rapid extraction of mineralized from hyperspectral remote sensing images using ensemble learning model has significance for resources exploration. In this study, two minerals, limonite chlorite, exposed at the surface Qinghai Gouli area were used as research objects. Sparrow search algorithm (SSA) was combined with random forest (RF) gradient boosting decision tree (GBDT) models, respectively, to construct indicative information models in study area. Youden index (YD) ore deposit coincidence (ODC) applied evaluate performance different extraction. The results indicate that optimization SSA parameter is obvious, accuracy both integrated after been improved substantially, among which SSA-GBDT best performance, YD ODC can reach 0.661 0.727, respectively. Compared traditional machine model, higher reliability stronger generalization application, greater than 0.6. addition, distribution extracted by basically consistent pattern fracture tectonic spreading characteristics known deposits (points) area, line mineralization Therefore, classification based on technology, efficient method.