Research On Landslide Crack Monitoring Technology Based On Close Range Image Recognition

China is a country with frequent landslides,which often pose a serious threat to people's lives and property.Therefore,our country attaches great importance to the study of landslide monitoring.The change of landslide crack is an important index of landslide,so it is of great significance for the research of landslide crack monitoring method.In this paper,a high-precision and real-time monitoring scheme for landslide crack monitoring is proposed.In this scheme,the close shot mode of monocular camera and depth camera is mainly used.Through real-time image recognition,the three-dimensional coordinates of the center point of the characteristic figure set in advance in the world coordinate system are obtained,and the distance changes between the center points of each figure are calculated,so as to dynamically monitor the width changes of the landslide cracks.The specific research work is as follows:1)Aiming at the image noise and the region to be processed,this paper studies and analyzes four common and practical image preprocessing technologies.They are image enhancement,image thresholding,edge detection and image filtering,and each preprocessing method is compared and analyzed.2)In the face of complex surrounding environment of landslide cracks,eliminating the influence of background information is the main means to improve the monitoring accuracy.In order to achieve this goal,fixing special marker points is a common means to eliminate the influence,among which the detection of circular marker points is a simple and practical method.In Chapter 3,the common algorithms of circular detection are introduced,including Hough transform,roundness detection and template matching Matching algorithm.In addition,the above circle detection algorithms are compared by experiments.The accuracy of circle detection algorithm and Hough transform is verified by experiments.The pixel coordinates of circle center are obtained by Hough transform algorithm.3)Based on the circle recognition algorithm and the principle of monocular camera imaging,a scheme of landslide crack monitoring is proposed in this paper.In this scheme,the preset circle label is placed on both sides of the landslide crack and fixed.The width change of the landslide crack is reflected by detecting the distance change between the circle marks.The experimental results show that the method is feasible,and the accuracy of the test is stable within 2 mm.In addition,the influence of camera translation and camera shooting angle change on the experimental accuracy is also explored.The experimental results show that the change of camera position will not affect the experimental accuracy.