Feature-based UAV River And Lake Image Intelligent Processing System

River and lake management is an essential and important link in the process of urban ecological management,which is related closely to residents' life and work.However,due to the insufficient application of science and technology in management work,most front-line work is done by hand,the huge workload leads to the contradiction of high cost,long cycle and low rate of return of river and lake management becoming increasingly prominent.Therefore,An intelligent processing system of rivers and lakes' images by UAV based on feature is proposed in this thesis.The system will apply computer vision technology in river and lake inspection to realize the transformation from artificial identification to machine identification,and to provide digital and intelligent technical support for the inspection of rivers and lakes.Firstly,this thesis analyzes the application status of UAV and the research status of AI at home and abroad,collects the application cases of UAV in water industry in China.Then according to the application requirements of rivers and lakes' images intelligent processing,studies the common techniques in image processing,and mainly solves two aspects of the problem:river segmentation and identification of two kinds of river surface pollutants,in particular,abandoned ships and green floats pollution.In order to solve the problem of river region segmentation,a segmentation method based on region growth and active contour model is proposed in this thesis,and experiments show that this method is more suitable than edge-based method for river region segmentation in rivers and lakes' images by UAV.To solve the problem of river pollution identification,DPM model and HSV model are used to identify abandoned ships and green floats in this thesis.Experiments show that DPM model is effective in identifying closely connected ships with different shapes and types.The accuracy of template matching method based on HSV can reach 91.23%in green floats recognition.Finally,based on the research of Jiangyan river and lake patrol workflow,this thesis carries out the requirement analysis and function design of river and lake image intelligent processing system,and presents the main functions of system such as UAV ground data acquisition,UAV data processing,River and lake inspection event processing and intelligent analysis of large data.This system realizes the intelligent processing of rivers and lakes' images,the digitized analysis of problem results and the automatic circulation of management work.It can reduce the cost of inspection,shorten the inspection cycle and improve the efficiency of inspection.It has good application prospects in the maintenance of river and lake health,the realization of sustainable utilization of river and lake functions,the improvement of water management system and the guarantee of water safety.