Research On UAV Route Planning And 3D Modeling Application For Debris Flow Ditch

Debris flow disasters occur frequently in Dongchuan,Yunnan.The occurrence of debris flow disasters has the characteristics of sudden,gradual,sequential,and concomitant nature,which makes it more difficult and risky to interpret and analyze the degree of disaster and carry out post-disaster work.At present,Unmanned Aerial vehicle-remote Sensing(UAV-RS)technology is widely applied to the investigation and monitoring of various geological disasters due to its advantages of flexibility,mobility and high image resolution.The areas where debris flow disasters occur are mostly plateaus,mountains and canyons,with large vertical drops,unstable air currents,and complex terrain and terrain.In the use of UAV for route planning and data acquisition,there are reduced flight safety,poor image quality,Long flight time,etc.At the same time,in the denoising processing of 3D modeling point cloud data,there are problems such as large data volume,poor filtering effect,and low overall adaptability.Therefore,it is the key to carry out debris flow disaster investigation and monitoring in the debris flow gully area to study how to select an appropriate route planning scheme for data collection and select a highly adaptive filtering algorithm for 3D modeling point cloud denoising processing.Taking the Dabai Nihe Gully in Dongchuan District of Yunnan Province as the research area,this paper carried out the research on the UAV route planning scheme and the evaluation analysis method for the UAV route planning for different disaster types in the debris flow valley,and the point cloud extraction for complex 3D modeling in mountainous areas,Filtering and denoising processing analysis and research,and finally realize the application of debris flow disaster monitoring under the best data results.The main conclusions of the paper are as follows:(1)Aiming at the three common disaster types of debris flow ditch,collapse,landslide and debris flow gully,the elevation route and debris flow gully video route planning for collapse and landslide are proposed,and the proposed route planning scheme is concrete and feasible.(2)Propose an evaluation and analysis method for UAV route planning,focusing on image data quality and 3D model quality,combining the four angles of image overlap,flight height difference,3D model detail texture,and model accuracy to carry out different route planning plans evaluation and analysis.Based on Matlab to realize the quality analysis of related image data such as image overlap,altitude difference,UAV route map,etc.,verify the constructed program through the existing three different types of features(urban buildings,mountain landslides and debris flow valleys).Experiments have proved that the method of using the Scale-Invariant Feature Transform(SIFT)and Random Sample Consensus(RANSAC)algorithms to optimize the homography transformation matrix is used to calculate the accuracy of image overlap and the original accuracy index error Within 10%,the accuracy is high,other analysis methods can be realized and the applicability is high.(3)The detailed texture and model accuracy analysis of the 3D model shows that for collapse and landslide,the elevation route is adopted.The route overlap is 98.72%,and the side overlap is 88.82%,which is better than the overlap of single horizontal and crossed routes.The detailed texture information of the 3D model is kept intact,and the accuracy of the model is better than conventional route planning.For debris flow valleys,video routes are used,and the average overlap is 96.12%,which is better than the overlap of single-level and tic-tac-toe crossing routes,but the 3D model has poor detail texture and serious blurring,and the model accuracy is lower than that of the tic-tac-toe crossing route planning.The zigzag crossing route is more suitable for debris flow ditch.(4)Analysis of 3D modeling point cloud thinning and filtering denoising process shows that,using of voxel grid filtering algorithm thinning comparison experiment shows that when the grid distance is 0.8m,the amount of data can be reduced and the point cloud details are obvious.The paper proposes a combination of voxel grid filtering,through filtering,cloth simulation filtering,and progressive morphological filtering.The filtering algorithm combination optimization according to the elevation threshold processing,the filtering effect is superior to the conventional method,the adaptability is high,and it has certain practical application value.(5)The types of disaster bodies in the debris flow valley area are identified and the spatial distribution characteristics are analyzed.The area can be divided into disasterprone areas,light-hazard areas and stable areas,which are obtained from the horizontal and vertical profile analysis of the divided areas.The area with more disaster bodies is in the shape of a "V",the central flat area gradually widens to the stable area,the slopes of the mountains on both sides become slower,the stable area is vertical and horizontal,and the erosion marks are obvious.The slope of the landslide body is generally greater than 32°,the slope of the landslide gully is concentrated in 16°-40°,and the two slopes are distributed in the east and west directions.(6)Obtained by statistics of the length and area change of the landslide gully boundary,the length of the gully boundary will increase in 2020,but its area will decrease;the normalized elevation index(NDEI)is used to calculate the difference between the two-phase elevation data,Using elevation,profile line,slope and aspect to analyze the changes in topographical characteristics,the landslide gully has a landslide phenomenon during 2019-2020,and the overall trend is downward.The overall change is mainly in the elevation of the middle of the gully,close to the gully.The elevation of the marginal area has decreased.The elevation of the two periods of data has not changed much,but the surface undulations in 2020 have changed significantly.