Optimization Of UAV Photogrammetry Path Planning Algorithm For Mountainous Areas

As an important means of production,land plays an important role in human production and life.Land engineering can improve the quality of agricultural land and increase the area of cultivated land,thereby increasing the degree of land-saving and intensive use,and further improving agricultural production conditions,effectively alleviating the shortage of land resources.Land engineering projects require long-term dynamic monitoring during the preconstruction period,during the construction process,and after completion,but traditional surveying and mapping techniques and surveying instruments are inefficient and timeconsuming,and it is difficult to meet the rapid development of land engineering monitoring needs for surveying and mapping science and technology.With the continuous progress and development of surveying and mapping hardware technology,unmanned aerial vehicle platforms and lightweight sensors have been widely used in the field of surveying and mapping.Unmanned Aircraft System Remote Sensing(UAVRS)technology has also begun to emerge and achieved rapid development.Compared with traditional surveying and mapping technology,UAV remote sensing technology has high surveying accuracy and surveying efficiency.The application of this technology in the field of land engineering has greatly improved the work efficiency of land remediation projects.However,UAV photogrammetry is greatly restricted by flying height,especially when it is used in research areas with large terrain fluctuations.This will cause a partial overlap rate to be lost between two consecutive images,and some areas cannot be repeatedly observed,thereby affecting data quality.In order to avoid insufficient coverage,the commonly used method in practical applications is to set the image overlap rate much higher than required,but this means that the distribution of waypoints will be denser,further causing data redundancy.In addition,due to the limited life time of the drone,the battery-powered drone can only support 15 to 30 minutes of flight,which is not enough to complete the photogrammetry task at one time.Based on the above reasons,this study attempts to develop an algorithm that can optimize the flight path of the UAV when performing photogrammetry tasks,and obtain the high-quality digital model in the mountainous hilly area with the lowest data amount and the shortest flight time.Product image collection.The Digital Surface Model(DSM),user-defined GSD and image overlap ratio in the study area will be used as the initial input of the algorithm.Based on the digital surface model,the algorithm can calculate the position of all waypoints in the flight mission according to the GSD and image overlap required by the user.Then,the algorithm will delete those redundant images to find the minimum number of images that completely cover the study area.Finally,the algorithm is used to design a path with the shortest flight time.The algorithm will eventually output a path with the lowest time-of-flight cost,which is composed of the positions of all waypoints used to obtain images and their access sequence.This study focuses on the problem of path planning for drone photogrammetry in mountainous and hilly areas,and proposes a statistical algorithm based on the number of times the target features are observed at the pixel level The conventional "S" shaped flight path is measured and optimized using the simulated annealing algorithm to achieve the planning of the shortest flight time path.The conclusion of the study is as follows:(1)Convert the research problem into an algorithm problem and split it into three parts: dense waypoint network layout,redundant data judgment and deletion,and minimum flight time planning.Corresponding algorithm ideas are proposed for different functions of each part.MATLAB is used to complete the code implementation of the algorithm,and five different scenarios are tested to test the performance of the algorithm in different environments.(2)The DSM model,camera parameters,image overlap rate and GSD of the study area are used to output the dense waypoint network.Using the dense waypoint network as input,the coverage of the images taken by each waypoint was determined.And use the cell array to store the coverage of each image,and then count the number of times each pixel on the target feature is observed.Use the while loop in MATLAB to realize the repeated selection of all images and determine whether it meets the deletion conditions,and terminate the loop after not deleting the images for 300 consecutive times.The output at this stage is the minimum number of waypoints after screening to ensure complete coverage of the study area.Then,using the selected waypoint network as input parameters,the flight path planning is simplified as a TSP problem,and the simulated annealing algorithm is used to optimize the flight path,and finally the path with the shortest flight time is output.(3)Tested the performance of the algorithm under different terrain models and different input parameters.The research shows that under the same “n” value,the data volume of the waypoint network optimized by the algorithm is greatly reduced,with a minimum of 23.51% and a maximum of 80.10%.The amount of data reduction depends mainly on the image overlap rate set when formulating the drone photogrammetry mission.The higher the overlap rate,the more obvious the data optimization effect.The maximum reduction in flight time is 5.82% and the minimum is 0.The improvement effect is not obvious mainly because the "S" path as the initial solution is a scientifically ordered path in a certain sense.The test results show that the algorithm proposed in this study can provide the optimal path with the minimum number of waypoints and the shortest flight time that completely cover the study area under different test scenarios.Especially when studying the flight path planning of mountainous hills,it has a good planning effect.