Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Master → Chine → 2019 → Study on Surface Micro-plaque Identification of Desert Steppe Based on Hyperspectral Image

Inner Mongolia Agricultural University (2019)

Study on Surface Micro-plaque Identification of Desert Steppe Based on Hyperspectral Image

朱相兵

Titre : Study on Surface Micro-plaque Identification of Desert Steppe Based on Hyperspectral Image

Auteur : 朱相兵

Grade : Master’s Theses 2019

Université : Inner Mongolia Agricultural University

Résumé
The micro-plaques on the grassland surface mainly include bare ground,rat holes and sand spots.Bare land is the starting point for grassland soil degradation,and the next step will be degraded into patchy sand spots,which are important factors that exacerbate the process of grassland degradation.Therefore,the national standards are based on the ratio of bare land area and rat hole area to the total area of grassland.However,the acquisition of this data has continued the method of manual investigation for many years,which is time-consuming and laborious,and cannot meet the needs of large-scale grassland degradation and rodent monitoring.The study area selects the typical desert steppe in Wulanchabu City,Inner Mongolia Autonomous Region,collects hyperspectral ground data under natural light,and uses the spectral index threshold method to explore the spectral information distribution interval and the optimal segmentation threshold of micro-plaques,and then the grassland surface.The micro plaque is highly accurate.In this paper,the surface micro-plaques of desertified grassland are taken as research objects.The ground hyperspectral data is analyzed and processed by using three kinds of vegetation indices(RVI,NDVI,SAVI),and the spectral distribution of micro-plaques in the test area is counted by threshold method.Analysis,found that the above three indexes can not effectively identify the rat hole and bare soil.This study proposes and applies for the first time the Micro-plaque Index Threshold(MPI-T)based on hyperspectral desert steppe,which overcomes the classification accuracy caused by the randomness of grassland land distribution and the complexity of image background.Low problem.The results show that the micro-plaque index threshold(MPI-T)method can make the overall classification accuracy of grassland surface micro-plaques reach 92.6%,and the Kappa coefficient reaches 0.889,which meets the hyperspectral recognition requirements of grassland surface micro-plaques.The micro-plaque index threshold method(MPI-T)proposed in this study provides a preliminary method and basis for the monitoring and prevention of grassland rodent damage using hyperspectral remote sensing,as well as grassland degradation monitoring and quantitative inversion

Mots clés : Hyperspectral; Desertification grassland; Surface micro-plaque; Vegetation index; Threshold method;

Présentation (CNKI)

Page publiée le 1er mai 2020