【作者】 王少军；

【导师】 薛重生；

【作者基本信息】 中国地质大学 ， 地图制图学与地理信息工程， 2013， 博士

【摘要】 论文选择中国矿山遥感调查理论与方法研究是国家目标和公务管理平台建设的急需。论文的研究方向是现代遥感高新技术应用理论和技术方法。但从矿山管理的整体性来看,本选题的研究风险性是遥感技术应用与国家法规标准的对接问题。由于中国矿山管理具有部门的多重性和管理上的叠加性,国家—地方—企业三者之间利益的矛盾是客观存在的,这就给矿山遥感调查与监测派生出更高层次的难题。论文研究试图避开一些政治和法律问题,重点研究中国矿山遥感调查与监测中技术层面的科学问题。本文虽然涉及到国家法律和法规方面的问题,但需要强调的是论文的核心问题属于自然科学领域。矿产资源开发是中国工业化的基础,中国的工业化进程是人类历史上参与人口最多、规模最大的工业发展过程。工业化带来的是资源过量消费和环境过度负荷两个问题。20世纪80年代后期,我国矿产资源开发秩序比较混乱、矿山环境破坏十分严重。在1995年、2005年国务院两次下文在全国范围内进行矿产资源开发秩序整顿工作,国家目标是保持良好的矿产资源开采秩序,减少矿山地质灾害、矿难事件发生。在近十多年来,国土资源部提出并在计划中设立“矿山多目标遥感调查与监测”项目,极力推进空间信息科学在我国矿山管理中的研究和应用。近年来,对地观测技术已经全面应用于地球表层的资源、环境及区域变化的数据获取与监测等领域。遥感信息技术为全球性的整体研究提供最新、最快、最精确的科学数据,为资源环境问题的监督管理提供了空间管理的技术平台。矿山遥感调查与监测的理论及方法就是针对“如何快速、准确、实时地发现违规开采目标和矿山环境问题”而展开的。作者系统总结了多年来在矿山遥感调查与监测工程中的矿山目标图像识别和检测的认知理论、图像理解、地表目标检测与动态变化信息提取、动态目标多层次识别与多时相变化信息拾取等前沿理论与关键技术问题。论文研究的理论基础是图像信息科学理论,研究涉及矿山地质理论、矿山地理理论、矿山生态环境理论、矿山灾害理论等领域。其科学思路是首先基于矿山地质—地理—地貌—生态—环境的信息建立矿山综合知识系统,然后依据综合知识体系构建从图像数据到矿山知识的图像标记转换体系。论文改变了传统的“看图识字”的遥感工作模式,在矿山遥感调查理论研究中,构建创新的知识体系；在图像信息转换方法研究中,构建其关键技术体系。论文基于图像信息转换及尺度转换理论,为矿产行政管理和矿山地质环境监测提供技术支撑,提出了基于综合知识结构的矿山目标图像标记理论,建立了矿山过程的地质目标—地貌目标—环境目标—矿山建筑物目标—地表人工堆积体目标—矿山污染目标等图像标志及目标拾取的方法体系。基于遥感图像的空间维、光谱维、时间维等多维信息,获取矿山目标图像成分的标记属性。图像标记是最原始的图像属性信息,具有一定的不确定性和模糊性,需进一步依据综合知识系统和空间结构特征进行知识挖掘处理和图像结构分析,减少图像标记的不确定性和模糊性的非确定性影响,建立服务于科学管理的图像标记属性及矿山执法依据的语义属性。论文基于矿山地理学及矿山地质环境调查与监测的研究框架,依据国土资源部计划项目提出的技术标准,构建了矿山多目标图像识别体系与遥感监测方法体系。包括：(1)矿山目标的标志要素体系：系统研究了矿山地质地貌环境的图像识别的知识体系,如地质结构、岩石光谱、地理区位、生态环境、地表覆被、开采类型、矿山属性、土地类型、水体类型、废渣堆积部位、废渣二次开采利用、坑道类型、矿坑排水范围及水污染类型等；(2)矿山目标的动态监测标志体系：利用多时相图像的光谱差异、形态差异、纹理差异、环境差异、人文差异等变化信息,基于遥感图像的光谱特性、尺度特性、时相变化特性、地理环境特性和交通设施特性等变化信息,实现了对矿山开发环境的动态监测和信息提取；(3)基于层次分析和主组分分析理论,运用图像时空信息重构方法,提取矿山地质环境及地质灾害动态信息。建立了单一时相矿山灾害调查和多时相灾害监测的预测方法；建立了矿山废弃堆积体、泥石流的危险性评价方法；建立了土地恢复现状调查评价方法。对各类动态监测及预测成果进行了现场检测验证。论文分六章,主要内容及章节如下：第一章绪论介绍了论文研究背景,评述了矿山遥感领域的文献,综合了前沿理论和主要研究方法；第二章重点论述了矿山目标图像认识理论及其应用成果；第三章论述矿山遥感图像识别机理和图像理解的各类推理、分析与提取方法；第四章论述了矿山环境效应与图像综合应用方法；第五章论述了矿山环境灾害遥感调查、监测和预报方法；第六章是论文的结论与展望。第二章至第五章为论文的重点和核心,属于作者独立研究成果。主要研究学术成果及创新点如下：1.建立了不同地区、不同类型矿山目标的图像识别理论体系,对推动矿山多目标遥感调查监测具有实践性意义。基于图像地学信息转换理论,构建了矿山遥感的区域地质地貌图像结构格架、含矿地层与含矿岩体的图像标记属性和综合性图像标志,解决了矿山目标在大区域尺度的空间定位问题；构建了矿山目标(包括矿山开采面、固体废弃物、矿山道路、矿山建筑物类型、选矿设备和尾矿库等)的空间结构要素及其地质地理环境组合信息的图像标记和综合识别标志。对矿山目标的空间定位、采矿权属边界复核及行政执法具有指导意义。2.基于循环经济学理论,引入矿产资源的“资源—产品—再生资源”的反馈式流程,依据矿山废弃物质的闭环流动时空特征,对矿山固体废弃物二次利用问题,提出了我国非常规矿产资源利用和开采的图像识别标志及解译评价方法。3.运用图像光谱维度、时间维度、空间维度及观测方向维度等综合信息,从不同角度建立矿山目标的识别标志及图像解析证据,对快速发现和评价矿山目标及其属性的语义化具有科学意义。本文提出将矿山目标宏观识别法和微观解析分析法相结合,完善了从遥感图像数据到矿山地理环境信息再到矿山地质环境监测信息工程的图像信息转换的理论方法体系。矿山目标信息来源具有多途径、多重性、非均匀性及非线性特性,基于层次分析和组分分析,对图像多重语义信息进行重构及转换,建立了矿山遥感人机交互式提取模式和流程。时空尺度转换问题也是矿山遥感调查图像认知的基础理论问题,论文就矿山遥感中的尺度效应、尺度转换等问题进行了深入探讨和研究。4.从矿产资源开采的地理过程及各阶段的地质地理环境效应出发,全面总结了矿山目标的时空演变特征的图像识别过程。论文系统研究了矿山开发过程中,不同类型矿产、开采方式对地形地貌、土壤环境、水环境、生态环境、景观美学环境等方面的影响。基于“近代人地关系学”理论,系统总结了矿山开发的环境影响因素及其时空耦合效应；提出了矿山地质环境问题对不同地质地理环境的响应模式的差异性。依据矿山环境效应存在空间上的超越性和时间上的滞后性,创新性地提出了矿山环境破坏的放大效应和延时滞后效应。初步构建了人—矿和谐发展的绿色矿山发展模式。5.在矿山地理学研究框架下,完善了矿山地质环境、生态地理环境、流域污染环境、土地复垦和生态环境恢复的遥感调查评价与监测的知识结构体系和方法体系。应用空间技术研究了赣东北德兴铜矿区固体废弃物产生泥石流可能危害的范围、重金属可能污染的区域；针对赣南定南稀土矿区废弃物相对松散、降雨量丰富的特点,建立了半定量泥石流预测模型,对可能产生泥石流的区位及强度进行预测,预测结果与实际相符合,为政府实施综合治理提供了科学决策依据。由于地理环境复杂、矿产资源类型多样化,地球圈层之间相互干扰,矿山目标分散等原因,基于遥感图像信息的地学图像理解和实际应用还存在着一些理论问题和关键技术问题需要进行探索性研究。(1)对于图像工程的三个层次(图像处理、图像分析和图像理解)的划分、定位与矿山目标识别结合紧密度不够,注重了图像分析和图像理解,图像认知学涉及较少。图像特征提取时从区域地学规律分析多,定量算法的实现较少。在后续研究中需要深化,为图像分析专家寻找新的结合点。(2)矿山环境地学特征复杂,管理上分属于不同职能部门,调查与评价体系有待完善。更多还原

【Abstract】 The research on the survey and monitoring theory and method to mine remote sense (M-RS) in China is necessary for the national goals and the construction on official management platform. Its research direction is in the application theory and the technique of the modern high-tech remote sensing field. However, viewed from the integrity of mine management, the study difficulty in this topic is the joint between the application on remote sensing techniques and the state law standard. Owing to the facts that the mine management in China is multiplicity and superposition and the benefit conflict among the State-the Local-the enterprise objectively exists, the more challenge to survey and monitoring on M-RS has been derived. On attempting to avoid some political and legal issues, this thesis is focused on the technical problems in the survey and monitoring on M-RS in China. In spite of the issues on laws and regulations involved, the essential question of this thesis belongs to the natural science domain.The mineral resource exploitation is the basis of China’s industrialization. The industrialization process is the industrial development whose participation is the most and whose size is the largest in history. The industrialization brought two regional issues on the excessive consumption on resource and the over-load on environment. In the late1980s, the mineral resource exploitation was disordered and the mine environment was seriously damaged in China. Therefore, in1995and2005, the State Council issued documents to rectify the mineral resources mining order for twice in1995and2005respectively. The national goal is to keep the well development order and reduce the geological hazard and the mine disaster. In recent years, the Ministry of Land and Resources put forward and established Multiple-Goal Survey and Monitoring on Mine Remote Sense and promote the research and application of spatial information science on mine management. Recently, the earth observation technology has been comprehensively applied in the data acquisition and the monitoring on the resource, the environment and the regional change of the surface. The remote sensing technology can provide the newest, fastest and most accurate scientific data for the global study and provide the technical platform of spatial management for the supervision on resource environment. The theory&method in survey and monitoring of M-RS will be developed aiming at how to discover the illegal mining objects and the mine environment problem rapidly, accurately and immediately.The author has summarized the leading theory and the key technology in the cognitive theory of the image recognition and detection on mine object, the image understanding, the detection and dynamic information extraction of surface object, the multilevel recognition and multi-temporal information of the dynamic object. The theoretical basis of this thesis is the image information science theory. In the domain of M-RS, this thesis involves the mine geological theory, the mine geographic theory, the mine environmental theory and the mine disaster. Its scientific idea is (ⅰ) to establish the synthesis knowledge system among mine geology-geography-geomorphology-ecology-environment and (ⅱ) construct the image symbol transformation system from the image to the mine knowledge according to the comprehensive mine knowledge system.The contribution is to change the original remote sensing pattern-Learning to Read by Pictures, build the innovative knowledge system in M-RS and establish the key technology system in the image information transformation study. Based on the theory of image information transition and scale transformation, this thesis may provide the technical support for the mineral administrative management and the monitoring on mine geological environment. The author proposes the theory in image symbel of mine object with the comprehensive knowledge structure and builds up the method system of the image symbol containing geological object-geomorphological object-environmental object-mine building object-artificial accumulation object on surface-mine pollution object and the information extraction. To obtain the label attribute on mine object image with the multi-dimension and multi-attribute information on remote sensing image such as the space dimension, the spectral dimension and the temporal dimension. The image symbol is the original image attribute information, with certain uncertainty and fuzziness, which is required to make the knowledge mining and picture structure analysis on the basis of the theoretical knowledge and the spatial structure of mine geology-geography-mining environment in order to minimize the uncertainty influence by the image uncertainty and fuzziness and define the image symbol attribute for scientific management and the sematization attribute for mine enforcement.Upon the mine geography and the framework of survey and monitoring on mine geological environment, according to the special technique standard of planning project design by Ministry of Land Resources, the author has built the image recognition system and the method system of remote sensing monitoring for the mine multiple objects, including (ⅰ) the symbol system:the author has studied on the knowledge system of image recognition for mine geological-geographical environment, i.e. geologic structure, rock spectrum, geographical location, environment, surface coverage, mining type, mine property, land type, water type, waste accumulation position, secondary waste recovery, gallery type, mine drainage scope and water contamination type;(ⅱ) the dynamic monitoring symbol system of mine objects:with the spectral difference, the morphological difference, the texture difference, the environmental difference and the cultural difference in the multi-temporal image, based on the spectral feature, the scale feature, the time-difference feature, the geographical environment feature and the transportation feature, the author has carried out the dynamic monitoring and the information extraction on the mining environment;(ⅲ) with the analytic hierarchy process and the principle component analysis, the author has extracted the dynamic information of the mine geological environment and the geological hazarder. In short, the author has prediction method mechanism for the single-temporal survey and the multi-temporal monitoring on mine disaster, the risk evaluation method to the mine waste accumulation and the debris flow, the evaluation method to the status survey on land restoration. The author has made the site check on all the dynamic monitoring and prediction results.This thesis is divided into six chapters and the main content is as following:Chapter1, in which the author introduced the study background, reviewed the literatures in M-RS and summarized the leading theories and main methods; Chapter2, where the author emphasized the image understanding theory on mine object and its application; Chapter3, where the author discussed all the inference, analysis on the image recognition mechanism, and the image extraction method for M-RS; Chapter4, where the author described the mine environmental effect and the comprehensive method to image application; Chapter5, where the author showed the method to remote sensing survey, monitoring and prediction for the mine environmental disaster; Chapter6, which is the conclusions and the prospection of the thesis. The key is from Chapter2to Chapter5, which is the achievement by the author’s independent study. The main academic achievements and the contribution are as following:1.The image recognition theory of mine object in different regions and different types has a practical significance on the promotion of Multiple-Objective Survey and Monitoring on Mine Remote Sense. On the image geo-information transformation theory, the author constructed the image structure of the regional geology and geomorphology, the image label attribute and the comprehensive image label of ore bed and ore rock body for M-RS and solved the spatial location problem of mine object in large scale; built the image label and the comprehensive recognition mark of the spatial structure element and the geological&geographic environmental information for mine object (including the mining surface, solid waste, the mine road, the mine building, the preparation equipment and the tailing pond). That has a guiding significance on the spatial location, the review on the exploitation right boundary and the administrative enforcement for mine object.2.The introduction of feedback process Resource-Production-Renewable Resources for mineral resources and the recycle model Low Development, High Use and Low Emissions is according to the circular economy and in accordance with the spatial and temporal feature of closed flow for the mine waste. And the image recognition symbol and the interpretation&evaluation method for the utilization&mining of unconventional mineral in China has been raised as well.3.The recognition symbol and the image resolution evidence have been found out from different views with the comprehensive image information including the spectral dimension, the temporal dimension and the spatial dimension, which has a scientific significance on the rapid discovery and evaluation on the mine object and the semantization of its attribute. That combing the macroscopic recognition with the microcosmic analysis on the mine object has perfected the theoretical method system from remote sensing image and the mine geographical environmental information to the mine geological environmental monitoring. For the multi-approach, multi-plicity, non-humogeneity and nonlinear characteristic of the mine object information, based on analytical hierarchy process and component analysis, through refactoring and transforming the image multi-semantic information. The model and flow of image information extraction in the human-machine interaction for M-RS has been established. Moreover, since the temporal-spatial transformation is another basic theory for the image understanding for M-RS survey, the scale effect and the scale transformation in M-RS have been made a deep discussion on.4.From the geographical process in mineral resources development and the geological-geographic environmental effect in all the stages, the image recognition process of the temporal-spatial evolution feature for mine object has been fully summarized. The influence of different mineral type and mining patterns on the topography, the soil environment, the water environment, the habitat environment and the landscape aesthetics environment in the mine exploitation has been made sturdy on. Upon the theory Modern Man-Earth Relationship, the author has made a systematic summary on the environmental factors and the temporal-spatial coupling effect of mine exploitation; and the difference between response models of different geological&geographic environment from the mine geological environmental problem. The amplification effect and the lagging effect of mine environmental damage has been creatively raised in accordance that the mine environmental damage effect is transcendental in space and lagging in time. And then the green mine development model was constructed for human-mine harmonious development.5.Under the framework of mine geography, the author has improved the knowledge structure system and the method system of remote sensing survey, evaluation and monitoring for the mine geological environment, ecological&geographic environment, valley pollution environment, land reclamation and ecological environmental restoration. The author has made a systematical study on the probable damage scope of the debris flow caused by the solid waste and the possibly polluted region by the heavy metal in Dexing Copper Mine; aiming at the characteristic that the waste is relatively loose and the rainfall is abundant in Dingnan Rare-Earth Mine in South Jiangxi, the author has made the semi-quantitative prediction model for debris flow to predict the location and the intensity. As the result, the prediction matched the practical situation. Therefore, it may provide the scientific decision basis for the government’s comprehensive treatment.Due to the complex geographical environment, the diverse mineral resources, the mutural interference among earth spheres and the loose mine objects, there are some theoretical issues and key technical problems to study in the geosciences image understanding and its practical application for remote sensing image.(ⅰ) The division and location of the three layers (i.e. the image processing, the image analysis and the image understanding) in the image engineering has not been much matched with the recognition of mine objects. The author has emphasized on the image analysis and the image understanding, while the image cognition has not been involved enough. The regional geo-regulation was used much while the quantitative algorithm was relatively little. In the future study, the quantitative method should be improved to make the new joint for the image analysis.(ⅱ) For the mine geo-environmental characteristic is complex and governed by different departments in management, the survey and evaluation system should be improved in the future.更多还原