The Method And Software Platform Of Dam Disease Detection Based On UAV Rapid Patrol Inspection

By 2020,the total length of dams in China has exceeded 312,000 kilometers.In the course of Long-term Service,due to the influence of factors such as water level fluctuation,geological disasters and complicated environment,the seepage of the dam is serious.These factors even lead to dam-break and bring a huge threat to the production and livelihood of people.It is of great scientific significance and engineering value to study the fast detection system which is suitable for the dam disease detection and maintenance.For Dam damage,leakage and other diseases detection,the traditional manual detection method,which is low efficiency and time-consuming,is still the main.The combination of Unmanned Aerial Vehicle(UAV)inspection and machine vision can greatly improve the speed of data acquisition and the accuracy of recognition.However,at present,the UAV detection system based on machine vision has some problems such as low automation of data acquisition,low precision of disease identification and slow detection speed.To solve the above problems,basing on the fast data collection ability of UAV,this study aims at realizing the rapid inspection of dam engineering diseases by combining theoretical analysis,model test and field test.Founding on the method of rapid identification of dam disease by machine vision,this research presents a fast recognition method of Dam crack based on improved Frangi filter and a leakage recognition method based on infrared image.Then these method are used to realized the rapid inspection of the dam damage and leakage.At the same time,combined with the actual detection requirements,a set of software platform of the dike UAV rapid inspection system is developed,and the above methods and software platform are verified through experiments.The main contents of this paper are as follows:1.In the term of rapid identification of dam cracks,this paper improves the traditional Frangi filtering algorithm and designs a visible light fracture identification method based on the improved Frangi filtering.Firstly,SURF classifier based on support vector machine was used to quickly classify the visible light image data collected by UAV,which is beneficial to quickly determine the dike cracks.Then,the dam crack identification method based on improved Frangi filter is used to quickly identify the dike crack.Relying on the disease discrimination model of traditional Frangi filtering,this method introduces function terms of image enhancement and noise suppression.This can improve Frangi filtering algorithm in the ability of noise suppression and the sensitivity of extracting the damage details of the dam.Finally,the calculation methods of crack length,maximum width and distribution area are raised in this paper and verified by experiments.2.In the aspect of dam leakage identification,this paper presents an algorithm for dam leakage recognition based on morphological reconstruction.Firstly,the MAD median filter is used to remove the outliers in infrared images.Secondly,the image is smoothed by the generalized morphological filter and superposed with the top-hat transform to reduce the noise and enhance the infrared image of the dam.Thirdly,by using the fifth weight function and logarithmic transformation,the traditional morphological reconstruction method is improved.At the same time,by using the temperature gradient information of the leakage region,morphology reconstruction is constrained.In this way,the traditional method based on morphological reconstruction is more accurate to identify the infrared leakage area of the dam.Finally,the method used in this paper is verified by experiments and the leakage area is calculated.3.According to the actual engineering requirements,a software platform for crack and leakage identification is developed,which is suitable for the dam UAV aerial vehicle inspection system.The experiment verified the effectiveness of the system in this paper by field test.