Research On Technologies For Identification Method Of Cyanobacteria Bloom Using UAV Remote Sensing Data

Water quality monitoring is a necessary preliminary to protect water resources.It is also a hot research in the field of environmental protection.Through the water quality monitoring,we can obtain the water information and protect the ecological balance which can promotes economic development and provide security for the people's life.Water pollution frequently causes cyanobacteria bloom.The artificial water quality monitoring and the satellite remote sensing monitoring methods are low efficiency,high cost,not flexible,not fine,etc.One of the current problems needed to resolve is that how to overcome the above shortcomings to investigate water quality efficiently and accurately.Starting from the application of Unmanned Aerial Vehicle(UAV)remote sensing technology and combining with the actual demand of water in cyanobacteria bloom monitoring analysis,this paper completes the applied research on the data acquisition,treatment and interpretation of UAV cyanobacteria bloom identification method.The main works of this paper are as follows:1.The present situation of UAV remote sensing research and the realistic requirement of cyanobacteria monitoring are analyzed.This paper expounds the UAV remote sensing technology feasibility analysis and research significance in cyanobacteria blooms recognition application.2.This paper briefly introduces the composition of UAV remote sensing system and illustrates the specific operation process of the UAV remote sensing images acquisition in detail.In view of the experimental area image acquisition,this paper analyzes the characteristics of UAV remote sensing image,summarizes the advantages of UAV remote sensing platform.3.A grid method is designed to control the number of interest points in the stage of SIFT feature extraction strategy.The major problems of UAV images mosaics are as follows.Images matching need take long time,occupy a large memory.Through the grid division,extreme of interest points are calculated,then we add constraints to reduce the threshold value in grid for the number of matching feature points.The fast matching of UAV remote sensing image will be realized.4.The enhanced red-green difference vegetation index is constructed,which is suitable for the optical image of the UAV.A reflection of RGB three bands are combined with algae in the water chlorophyll,and blue band reflectance characteristics.Through the processing of images band,the algae remain remarkably consistent in the water.The vegetation index improves the recognition rate of low concentration of cyanobacteria bloom.5.This paper establishes a model of the Support Vector Machine(SVM)after optimizing parameters.Based on the principle and method of support vector machine model,we select the radial basis kernel function which has a wide domain of convergence,then use the dual personality network to find the optimization parameters of the kernel function and carry out the cross validation to obtain the support vector machine model of the cyanobacteria classification.Finally this paper gets the classification results of water and cyanobacteria whose accuracy evaluation meets application requirements.