Extraction And Classification Of Vegetation Information In Coal Gangue Pile Based On UAV Remote Sensing

In the natural environment,coal gangue piles was prone to spontaneous combustion,leaching,dust and other natural disasters,causing serious pollution to the air,water and soil and so on.However,the natural restoration process of coal gangue piles was extremely slow,so it was of great significance to take effective artificial restoration measures for the ecological reconstruction of coal gangue piles for the surrounding ecological environment,people’s livelihood and economy.Among them,the effect evaluation of vegetation restoration was an essential link in ecological governance and restoration projects,which provided scientific basis and reference for the subsequent management,maintenance and similar governance and restoration projects of coal gangue piles.At present,the effect evaluation of vegetation restoration mainly relied on traditional methods such as field samples investigation,which not only consumed a lot of manpower,material resources and time cost,but also had relatively low efficiency.In this thesis,Changcun coal gangue pile of Luan mining area was taken as the study area.UAV remote sensing technology was used to obtain visible light image of the study area simply,quickly and at low cost.And the information of green vegetation in the study area was extracted with high precision and the information of dominant vegetation species in gangue pile was effectively identified.Finally,based on the results of optimal identification and classification,the effect of vegetation restoration in the study area was evaluated and analyzed from the perspectives of vegetation coverage and vegetation allocation pattern.The main research results and conclusions are as follows:(1)In this thesis,a new visible light vegetation index DEVI was proposed,which comprehensively utilized the red,green and blue wave information of visible images.And the green vegetation information in the image of the study area was extracted by the bimodal histogram threshold method and OTSU threshold method,respectively.The results showed that DEVI combined with bimodal histogram threshold method can gain optimal extraction result,and its overall accuracy was 98.18%,Kappa coefficient was 0.93,and relative error was 1/32,which was obviously better than other 14 common vegetation indices.At the same time,three typical vegetation coverage areas were selected to verify the feasibility of the index method,and the results showed that DEVI had good applicability and reliability.(2)Firstly,image enhancement techniques such as vegetation index,color space conversion and texture analysis were used to construct the classification feature information data set.Then,the improved artificial feature selection method and principal component analysis method were used to achieve dimensionality reduction of data set and the results were used as the basis of image classification.Finally,three machine learning classification algorithms were used to recognize and classify vegetation species information in the study area.The results showed that the recognition and classification accuracy of the improved artificial feature selection method combined with SVM classification algorithm was the highest,with the overall classification accuracy of 94.29% and the Kappa coefficient of 0.93.The accuracy of PCA combined with SVM classification algorithm was slightly reduced,with the overall classification accuracy of 88.35% and Kappa coefficient of 0.86.In general,the classification accuracy of vegetation information based on multi-feature was higher than RGB image.(3)Based on the results of vegetation information optimal identification and classification,the vegetation restoration effect of Changcun coal gangue pile was analyzed from the perspectives of vegetation coverage and vegetation configuration mode.The results showed that the vegetation coverage of the study area was 81.68%,and three vegetation configuration modes,including pure grass type,grass shrub type,arbor,shrub and grass type,were used for vegetation restoration.The plant diversity was rich,and the overall effect of vegetation restoration was good.