面向矿区大梯度形变监测的SAR影像Pixel-tracking关键问题研究

发布时间：2018-01-07 17:25

本文关键词：面向矿区大梯度形变监测的SAR影像Pixel-tracking关键问题研究　出处：《中国矿业大学》2017年博士论文　论文类型：学位论文

【摘要】：中国是煤炭生产和消费大国,大量的煤炭资源开采在为经济建设作出重要贡献的同时也带来一系列的环境和地质灾害问题。加强对矿区因煤炭资源开采引起的地表形变的监测对于矿区生态环境修复和灾害预防及管理具有重要意义。InSAR技术的出现为矿区形变监测提供了一种新的方法,然而由于受地表下沉盆地形变梯度的局限性,基于相位解缠的InSAR技术很难有效获取高强度开采过程中地表的大量级形变,只能对下沉盆地边缘小量级形变进行有效监测。基于SAR影像强度信息的Pixel-tracking方法为矿区大量级、大梯度形变的监测提供了一种新的技术手段。本论文重点研究基于SAR影像的矿区大梯度、大量级地表形变的精细化监测,其主要研究内容和成果总结归纳如下:(1)总结了现有InSAR技术监测矿区形变的研究现状,阐述了D-InSAR、时序InSAR及Pixel-tracking方法的基本原理,依据概率积分法模拟地表下沉,分析了不同波长、不同像元尺寸的TerraSAR-X影像、Radarsat-2影像和ALOS-PALSAR影像监测矿区形变的适用性。(2)提出一种局部自适应窗口的Pixel-tracking方法。利用概率积分法地表移动预计模型模拟不同地质采矿条件下引起的不同量级的地表形变,把模拟的形变加入三种不同像元尺寸的SAR影像强度信息中,研究了互相关窗口大小、像元尺寸大小、互相关系数内插因子等因素对Pixel-tracking方法监测精度的影响。发现过小的互相关窗口会造成像元误匹配现象,过大的互相关窗口会造成下沉盆地的“形变压缩”现象;对过大的互相关窗口造成下沉盆地“形变压缩”现象的原因进行分析,对SNR进行重新定义;依据矿区形变特征,对互相关系数峰值匹配位置进行约束,极大减少了像元误匹配出现的概率;基于重新定义的SNR,通过在一定的区间内变换窗口步长寻找SNR最大值,SNR最大值对应互相关窗口即为最优互相关窗口。基于最优互相关窗口对矿区形变进行监测,通过采用覆盖大柳塔矿区52304工作面Radarsat-2影像进行验证,证明该方法相对于固定互相关计算窗口的Pixel-tracking方法能够明显提高监测精度。(3)提出一种顾及地形影响因素改正的Pixel-tracking方法,削弱地形因素对Pixel-tracking监测精度的影响。利用时间序列SAR影像强度和标准差双阈值约束的方法对SAR影像中影像强度保持稳定的点进行初步筛选,按照均匀分布原则和避免采空区影响原则,对初选稳定点进行精炼,根据精炼后的稳定点下沉信息,采用最小二乘多项式曲面拟合方法对Pixel-tracking方法监测结果中的轨道误差、大气误差及部分地形误差进行削弱;依据稳定点的残差,引入外部DEM数据,基于高程信息进行二次多项式拟合,削弱地形因素对Pixel-tracking方法监测精度的影响,并采用52304工作面地表实测数据进行验证。(4)采用SBAS-InSAR思想对Pixel-tracking方法的监测结果按照一定的时、空基线约束进行组合,采用观测时段的形变量代替平均形变速率进行解算,获得每个观测时段的最优解,结合工作面开采掘进速度,对小基线集Pixel-tracking方法得到的时序形变结果进行分析,并结合地表移动观测站GPS数据评价了时序小基线集Pixel-tracking方法的监测精度。(5)优化多平台联合解算获取三维形变的模型,利用两个平台SAR数据可获取矿区三维形变场。根据Pixel-tracking方法既可以监测卫星视线向形变又可以获取方位向形变的特点,采用TerraSAR-X卫星和Radarsat-2卫星两个不同平台的时序SAR影像Pixel-tracking监测结果进行联合解算,获得52304工作面下沉盆地三维形变场,并根据地表移动观测站实测数据,对基于Pixel-tracking方法联合解算获取的三维形变场监测精度进行了评定。(6)提出一种先验定权的时序InSAR与Pixel-tracking监测结果融合方法。采用四种时间序列InSAR方法对52304工作面地表下沉盆地边缘小量级形变进行监测,并依据地表移动观测站GPS数据对四种时序InSAR方法矿区小量级形变监测精度进行评价,获得矿区高精度小量级地表形变监测结果;采用时序小基线集Pixel-tracking技术对52304工作面大量级、大梯度形变进行监测,依据先验方差定权对时序InSAR方法的监测结果和时序Pixel-tracking的监测结果进行融合,并根据融合结果获取开采沉陷部分参数。
[Abstract]:Chinese is a coal production and consumption, make an important contribution to the economic construction at the same time also brought a series of environmental and geological disasters in mining a lot of coal resources. To strengthen the monitoring of surface deformation caused by mining coal mining in the mining area ecological environment restoration and disaster prevention and management has an important significance of.InSAR technology as the deformation monitoring in mining area provides a new method, however, due to limitations of the surface subsidence basin terrain gradient, phase unwrapping InSAR technology is difficult to effectively obtain high strength mining in large deformation based on table level, only the effective monitoring of subsidence basin edge small deformation method based on Pixel-tracking SAR. The image intensity information for a large number of mining area, provides a new method for monitoring large deformation gradient. This paper focuses on the study of images based on SAR The mining area gradient, precise monitoring of large deformation, the main research contents and achievements are summarized as follows: (1) summarizes the existing research situation of InSAR technology in deformation monitoring, expounds the basic principle of D-InSAR sequence of InSAR and Pixel-tracking method, based on probability integral method to simulate the surface subsidence, analysis of different wavelengths and different TerraSAR-X image pixel size, the applicability of Radarsat-2 image and ALOS-PALSAR image deformation monitoring in mining area. (2) proposed a method of Pixel-tracking local adaptive window. Using probability integral method of surface movement prediction model by simulation under different geological and mining conditions of different magnitude of surface deformation, the deformation simulation of SAR image intensity information added three different pixel size, the correlation window size, pixel size, correlation coefficient interpolation factor and other factors on the Pixe Effect of l-tracking method for monitoring accuracy. Found correlation with small window will cause the pixel mismatch phenomenon, large correlation window will cause the sinking basin "compression deformation" phenomenon; cause analysis reasons of subsidence basin "deformation compression" phenomenon of correlation window, the SNR was re defined; based on the deformation characteristics of mining area, on the mutual relations between the peak position of the matching constraints, greatly reducing the pixel matching error probability; definition based on SNR, through within a certain range transform window step to seek the maximum value of SNR, SNR corresponding to the maximum cross-correlation window is the optimal correlation window. The optimal correlation window based on the deformation monitoring of the mine, by covering Daliuta mining area to verify the Radarsat-2 52304 working face images show that this method, compared with the fixed cross correlation window Pixel- The tracking method can significantly improve the monitoring accuracy. (3) proposed a Pixel-tracking correction method of terrain factors, weaken the effect of terrain factors on Pixel-tracking monitoring accuracy. Using time series SAR image intensity and standard deviation method of double threshold constraint on SAR image image strength to maintain the stability of the point of preliminary screening, according to the principle of and avoid the influence of goaf on the primary principle of uniform distribution, stable point of refining, according to the information sink stable point after refining the orbit error by using the least square polynomial surface fitting method of Pixel-tracking method in the atmospheric monitoring results, error and part of the terrain error are weakened; based on residual stable point, the introduction of external DEM data, elevation two polynomial fitting based on information, weaken the effect of topographic factors on the accuracy of the monitoring method of Pixel-tracking, and uses 52304 As the surface of measured data are verified. (4) by the method of SBAS-InSAR monitoring on the result of the Pixel-tracking method in a certain space, baseline constraint combination, the observation period shape variables instead of the average deformation rate calculation, obtain the optimal solution of each observation period, combined with the working face mining tunneling speed, timing of deformation the results obtained by Pixel-tracking method on small baseline subset analysis, combined with the monitoring precision of surface movement observation station GPS timing data evaluation of small baseline subset of Pixel-tracking method. (5) the optimization of multi platform combined solution to obtain the 3D deformation model can obtain 3D deformation field using two SAR data mining platform. Based on Pixel-tracking method can monitor the satellite line of sight to the deformation and can obtain the azimuth deformation characteristics, using TerraSAR-X and Radarsat-2 satellite platform of the two different time series S The monitoring results of image Pixel-tracking AR combined solution, 52304 working surface subsidence basin 3D deformation field, and on the basis of surface movement observation data, 3D deformation field monitoring accuracy of Pixel-tracking method based on the combined solution obtained was evaluated. (6) proposed a priori right timing InSAR and Pixel-tracking monitoring results fusion method. Using four kinds of time series InSAR surface of 52304 working surface subsidence basin edge small deformation monitoring, and on the basis of four sequential InSAR method of mine small deformation monitoring precision of surface movement observation station evaluation of GPS data mining, to obtain high precision small deformation monitoring results; the small baseline subset of Pixel-tracking sequence in 52304 working face of large scale, high gradient deformation monitoring, based on the monitoring results of prior variance right timing of InSAR method The monitoring results of the time series Pixel-tracking are fused and the parameters of the mining subsidence are obtained according to the fusion results.

【学位授予单位】：中国矿业大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：TD325.4

【相似文献】