Research On Binocular Visual Monitoring And Early Warning Methods Of Slope Surface Deformation

China is a mountainous country which mountainous area is about 1/3 of the land area,especially in the Midwest.With the further development of China's modern infrastructure construction,the center of construction has gradually shifted to the central and western regions.Therefore,a large number of slope projects needs to be faced in in the construction.Landslide,as a large-scale geotechnical disaster,has become an important factor affecting the safety of social life and property,regional economy and stability.The slope deformation information is a direct reflection of its disability,so it is key to improve the slope monitoring system to guarantee the safety of social life and property.At present,there are some problems in slope monitoring,such as low intelligence,discontinuity and low data quantity,so it is difficult to provide accurate deformation information of slope.The landslide warning is mainly based on the deformation information of key positions,so it is difficult to identify slip patterns,the method of threshold warning of measured data is also difficult to reserve enough time for landslide prevention and control,the difficulty of single early warning index in accurate early warning also exists.Therefore,it is necessary to further develop slope deformation monitoring and early warning technology.This paper aims to achieve effective slope deformation monitoring and accurate early warning.The high precision,high adaptability,intelligent deformation monitoring of slope surface are realized by binocular vision technology,and the accurate and effective landslide warning method based on monitoring data is established.The main research contents and methods include the following aspects:(1)The principle of binocular vision measurement is introduced in this paper.Based on the principle,the transformation process from image coordinates to spatial coordinates is completed,the binocular vision measurement model is constructed,and the slope surface deformation monitoring system is constructed.(2)The design of measuring point is studied and the design scheme of concentric circle target is put forward in this paper.The proposed high precision concentric circles image positioning method includes three steps:subpixel edge detection,elliptical fitting and center clustering.The stability of the method is tested,which shows that the algorithm has high precision and strong stability.The parameters of target are determined by simulation experiment.(3)The image acquisition of slope monitoring is studied.It shows that factors affecting image quality are fog,light,rain and noise.Image restoration technology was used to optimize the images in various environments,a slope monitoring image optimization system is established as well.Hope to eliminate the environmental impact to the greatest extent,and Improve the environmental adaptability of computer vision measurement technology.(4)The quantitative method of measuring point space displacement is studied,which consists of five steps:image distortion correction,stereo matching,target tracking,image positioning and space coordinate solution.This part mainly aims at improving the problem of the lack of accuracy of the existing target tracking technology.The high precision target tracking is realized,and the process of the target space position change is obtained.(5)A landslide warning method based on measured deformation data is presented.This method includes three main contents:deformation field reconstruction,displacement trend prediction and multi-indicator landslide warning.(6)The static and dynamic measurement accuracy of binocular vision system is analyzed through laboratory experiments.Based on the system,the slope surface deformation monitoring scheme is designed.In this study,binocular vision technology is introduced to monitor the deformation of slope surface;the concentric circle target is designed to improve the positioning accuracy of target;the monitoring image optimization method is designed;the slope monitoring image optimization platform is formed and the image acquisition of high quality slope is obtained.Based on binocular vision technology and improved target tracking technology,the change process of target space displacement is quantified.A landslide warning method is provided based on the monitoring data.It provides methods and ideas for deformation monitoring and early warning of slope surface.