Extraction of metamorphic minerals by multi-scale segmentation combined with random forest
- Available Online:
Abstract: To improve the extraction accuracy of metamorphic minerals from remote sensing images and further improve the recognition effect of metamorphic zones, we combines ratio operation, multi-scale segmentation and random forest classification to extract metamorphic mineral information from Aster images in Beishan area in Gansu Province. Initially, the image is enhanced by the ratio formula of the characteristic spectral structure of the target mineral; then, the image is segmented by multi-scale based on the spectrum and variogram; finally, the accuracy is evaluated by the thin film identification results of the field exploration samples after the target mineral is extracted by random forest. The results show that biotite, muscovite and amphibole have identification characteristics on ASTER image, and the extraction accuracy is 85.4088%, 84.7640% and 85.7308% respectively; the extraction accuracy of other metamorphic minerals with less content can reach more than 60%. Multi-scale segmentation can make full use of the clustering features of minerals; variogram texture can enhance the ability of morphological features to distinguish minerals; random forest is not sensitive to noise and the extraction results are stable.