Research On Damage Detection Of Urban River Revetment Based On UAV Photogrammetry

The damage detection of river revetment is of great significance to the management and maintenance of urban river bank protection.However,most of the damage detection work of river revetment still relies on manual visual inspection,which leads to high cost,long cycle and low efficiency of revetment management.Different from on-site visual inspection,UAV low altitude photography can completely scan the revetment surface without manual on-site visual inspection.Therefore,this paper focuses on the research of automatic identification method of damage characteristics of UAV aerial images,and applies low altitude UAV photogrammetry technology to Revetment detection,improves the efficiency of revetment damage detection,realizes the transformation from manual identification to automatic identification,and provides digital and intelligent technical support for revetment management.According to the management requirements of urban river revetment,this paper analyzes the research status of UAV low altitude Photogrammetry in urban management and target detection at home and abroad,and analyzes the commonly used technical methods.It is found that there are two aspects to be improved: one is to generate more high-quality dense point cloud of river revetment;the other is to accurately identify and detect the damage characteristics.The main research contents are as follows:(1)In order to generate higher quality dense point cloud,this paper studies the sub Harris algorithm combining scale invariant feature transform(SIFT)and Harris operator to extract and match features;through the matching algorithm combining region growing method and semi global matching,dense point cloud is generated pixel by pixel,and the geometry and fine surface of revetment are accurately restored.Pixel by pixel dense point cloud matching can not only restore the fine surface of revetment,but also significantly improve the accuracy of revetment damage identification by enlarging the distance between damaged area and undamaged area.(2)The slope intensity map is generated by calculating the slope of the dense point cloud generated;the revetment area is automatically extracted from the slope intensity map by using the improved six dimensional simple linear iterative clustering algorithm(6d-slic);the damage regions of different sizes in the slope intensity image are accurately identified by using an adaptive multi-scale gradient operator.(3 Through the use of consumer UAV low altitude acquisition of a river bank slope image in Nanchang City,using the research method of this paper,the automatic damage identification performance evaluation is carried out.The results show that the method can identify the damage characteristics of revetment accurately,quickly and economically.Therefore,this method can be used as an effective method to identify the damage of urban river revetment.This research realizes the automatic process of the detection of urban river revetment,realizes the goal of digital analysis and intelligent processing,reduces the detection cost,and has the characteristics of short cycle and high efficiency.It has a good application prospect in detecting the health degree of revetment,realizing the permanent utilization of river and lake functions and optimizing the water treatment system.