The Design And Implementation Of River Course Inspection System With Unmanned Aerial Vehicle

With the rapid economic and social development of our country,the problem of water environment deterioration caused by water pollution and illegal construction of rivers has become increasingly prominent.River course inspection is the main way to find such problems,but manual patrols have disadvantages such as low efficiency,high cost,and long cycle time.In this thesis an intelligent digital image processing system was developed for river patrolling based on UAV,which realizes the functions of image mosaic and intelligent comparison of patrol images,closed-loop processing of problem flow,and intelligent analysis of inspection data.This system improves the efficiency of river patrol,reduces costs,and shortens the cycle.This thesis elaborates the significance of research on the river course inspection system.The current situation of inspection at home and abroad were analyzed,the optimal type of UAV for river course inspection was selected,and the advantages of the system were introduced.In order to meet the real-time requirements of the system,based on the analysis of the traditional SURF algorithm,in combination with the fact that this system doesn’t need high accuracy,dimensional reduction is used for reduce the works of compute and match feature vectors in order to accelerate the speed of operation,an improved SURF feature extraction and matching-QSURF algorithm is proposed.100 groups of river images were compared using Jiangyan District aerial photography,indicating that about 40%of the stitching processing time is saved by the QSURF algorithm compared to the traditional SURF algorithm.According to the design principle of the system,the architecture of the system is clarified,and the concept structure and logic structure of the database were designed.The system achieved functions such as mosaic and intelligent comparison between images,processing of inspection events,and can analyze inspection data.The system was tested in terms of functionality,reliability,and compatibility and so on.At last,rearview the work done,summarizes the shortcomings and needs to be improved,and points out the need for further research and improvement in the future.