中文摘要：

针对现有的大型露天矿边坡大变形监测方法存在的不足,提出了一种基于ASIFT图像特征匹配算法的滑坡位移监测方法。首先基于露天矿滑坡真实地形空间分布形态,设计了滑坡物理模型,并利用高分辨率数码相机获取模型滑动过程中的序列影像;然后利用ASIFT特征匹配算法对滑坡各阶段光学影像进行特征点提取与匹配,并利用RANSAC随机抽样一致算法剔除误匹配点;最后利用滑坡体运动特征矢量集模型计算出特征点空间矢量位移,从而实现了滑坡位移场标定与滑坡范围确定。研究表明：所提算法对于滑坡蠕变阶段的位移监测效果一般,但可有效监测滑坡中后期发生明显变形时的位移,该算法的特征点匹配数量及匹配正确率均明显优于SIFT算法,且位移矢量标定精度也得到明显提高,反映出该算法适合于大型滑坡后期发生大变形时的位移量监测。

英文摘要：

In order to solve the shortcomings of the existing methods in the deformation monitoring of large scale open-pit slope,a method for monitoring landslide displacement is proposed based on ASIFT （ Affine scale invariant feature transform） image feature matching algorithm is proposed. Firstly,the landslide physical model is designed based on the real terrain spatial distribution of open-pit mine landslide,and the sequence images of the model sliding process is obtained through high-resolu-tion digital camera;then,the feature points of the optical photos of landslide in different stages are extracted and matched by u-sing ASIFT feature matching algorithm,and the error matched points are dropped by using RANSAC random sampling algo-rithm;finally,the spatial displacement vectors of feature points are calculated by adopting the landslide motion feature vector set model,so the landslide displacement field is calibrated and the landslide range is determined. The study results show that：it is very difficult to monitor the landslide displacement in creep deformation stage by using ASIFT image feature matching al-gorithm,but it could be effectively applied to monitor the obvious landslide displacement in slowly moving stage,sliding stage and slowly crawling stage,the feature points matching number and matching accuracy of the ASIFT algorithm are both superior to the ones of SIFT （ Scale invariant feature transform） algorithm,to be specific,the the number of correct matching feature points of ASIFT algorithm is 3~5 times to the one of SIFT algorithm,and the matching accuracy is increased from 75% to 97%,besides that,the calibration precision of displacement vectors is also improved obviously. The above analysis results fur-ther that the method proposed in this paper is suitable for the displacement monitoring of large deformation in late stage of large scale landslide.