| As my country gradually attaches importance to the protection of the ecological environment,in order to protect the ecological environment,restore the damaged ecological environment of the mines,and build the abandoned mines into lucid waters and lush mountains,my country has gradually begun to attach importance to the ecological restoration and governance of the mines left in history.The detection and identification of legacy mines and the geological environment assessment of relevant administrative areas will help the local government to collect statistics on historical legacy mines and formulate relevant ecological environment policies.This study takes the Lianhuashan eco-tourism resort area as the research area.As an important eco-tourism resort area around Changchun,it has high requirements for the quality of the ecological environment.Therefore,the image recognition and geological environment evaluation of historical mines in this research area are sufficient.necessity.The research of this paper is mainly divided into two research parts: 1)remote sensing image recognition of historical mines in the study area;2)geological environment assessment of the study area.The research contents of two parts are related to each other,and the results of remote sensing image recognition are used as the basic data for mine geological environment assessment.The main contents are as follows:In the identification of remote sensing images,high-definition satellite images are used as the research data source,and two interpretation methods,computer image recognition and interpretation and artificial visual interpretation,are used,and the results are verified through field investigations.In the computer image recognition and interpretation work,due to the small number of high-definition remote sensing image samples available,a transfer learning image recognition algorithm was introduced to detect and identify the mining pits and mining slopes of the historical legacy mines in the study area.In this study,two sub-models in the transfer learning image recognition algorithm are used for comprehensive discrimination,namely the pre-training model and the far-domain transfer learning model.1)In the pre-training model,5 structures are introduced for comparison,namely VGG-16,VGG-19,ResNet-50,ResNet-101,ResNet-152,in the image recognition of mining pits and mining slopes In the task,it is proved that ResNet-152 in the pre-trained model works best for structure recognition.2)When using the far-field migration model to identify the mining pits and mining slopes of historical mines,use the UC Merced Land-Use image set,the WHU-RS19 remote sensing image set,the SIRI-WHU remote sensing image set and the Bijie landslide image set.As a source domain dataset for fardomain transfer learning.After comparative experimental analysis,the quality of the model is judged by four indicators: reconstruction loss,domain loss,classification loss and total loss.Through the analysis of the results,it is known that the SIRI-WHU remote sensing image set performs best in the task of identifying mining pits.,the Bijie landslide dataset performs best in the task of mining slope identification.It has been verified that although the accuracy of computer image recognition is slightly lower than that of human visual interpretation,it can still be used in remote sensing interpretation due to its simplicity and speed.Based on the results of remote sensing interpretation work,this study conducts regional and comprehensive evaluation of the geological environment of historical mines.In the zoning evaluation of the geological environment of the historical mines,according to the results of remote sensing interpretation and field investigation,the internal areas of the historical mines are divided into mining pits,mining slopes,mining industrial squares and solid mines.Waste accumulation,according to the geological environment problems(geological disasters,topographical landscape damage,land resource occupation and destruction)caused by previous mining activities in each mining area,the geological environment zoning evaluation of historical legacy mines is carried out in accordance with relevant norms,so as to obtain Accurate assessment results of geological environment zoning in historical mines.According to the collected geological data of the study area,6 influencing factors of the study area were selected to evaluate the geological environment background of the mine.Use the relevant geological environment background of the study area and the relevant results of image recognition to make statistics and analysis on the main geological environment problems of the historical mines,select the geological environment background of the historical mines,the morphological characteristics of the abandoned mines,the internal units of the abandoned mines,and the abandoned mines.The main ecological problems of the mines and the restoration and management of abandoned mines are evaluated by multiple evaluation indicators.The analytic hierarchy process is used to conduct a comprehensive evaluation of the geological environment according to the specifications in the "Specifications for the Investigation and Evaluation of Mine Geological Environment"(DD2014-05).The comprehensive evaluation results of historical abandoned mines are finally divided into three levels: serious,relatively serious,and general.In the evaluation results,among the 88 historical abandoned mines in this study,we can know that 45 of these mines have serious geological environmental problems.There are 26 more serious ones,and 17 ones with normal geological environment problems.According to the comprehensive evaluation results of the geological environment of the Lianhuashan Eco-tourism Resort,combined with the development plan of the Lianhuashan Eco-tourism Resort,the study area has been analyzed from geological disasters such as collapse and unstable slopes,restoration of topography and landscape,and restoration of land resources.Provide scientific and reasonable suggestions for the improvement of the geological environment. |
PDF Full Text Request