Research On DEM Processing Method For Irrigation Networks Contour Extraction

Rapid mapping of irrigation networks contour has important significance for the water-saving irrigation techniques,rational allocation water,and water supplying safety.According to the drawback that the low rate of remote sensing image recognition,difficult extraction and more human intervention in irrigation networks contour extraction currently.Based on the the RGB imag,digital elevation model(DEM)and slope data measured by UAV aerial photography,This paper adopted the object-oriented method,supervised classification method,topographic features method and revised hough transform to extract the irrigation networks contour.Since the slope can reduce the impact of roads and farmlands in the irrigation networks contour extraction,enhancing the topographic features of irrigation networks.all the results were compared with the manual digitizing,as is proved,the integrity of revised Hough transform was the best among the other three methods.this method is an effective exploration of realizing irrigation networks contour extraction.The main research contents and achievements are as follows.(1)Design and implement the irrigation networks contour extraction method based on terrain features.According to the interference factors exist in irrigation district such as highway,lake and house,the prior information was added to algorithm implementation,the data was preprocessed,DEM was converted to slope.Secondly,the Optimal Threshold that distinguishes irrigation networks and non irrigation networks was obtained by testing the data set.Finally,according to sampling points in field investigation,the breadth-first search method is used to search and save the irrigation networkd contour information.It is proved that the integrity of this method is 66.95% higher than the result of object-oriented method in a variety of data,lower than the result of supervised classification methods in a variety of data.At the same time,the error is the lowest,only only 34.72%.(2)Design and implement the irrigation canal networks contour extraction method of improved hough transform.In irrigation districts,the field of farmland,roads and canal networks exist the features of structured and linear.but basically there is no gradient change in the fields of farmland and roads.Slope conversion can filter out roads and farmland information.Based on the gradient data,this article adopted a series of steps such as graying.filtering and noise reduction.binarization.removal of noise using connected region.Due to the advantages of Hough Transform while extracing lines in much noise.This paper implemented the extracion of irrigation networks contour,and the same times,this method revised the issues of fracture and intersection.As is proved,the revised Hough Transform has the best integrity,which can reach 85.61%,and is higher 32% than Object-Oriented method of slope data.the error rate is lower than 47.33%。at the same times,the integrity is higher 3.77% than supervised classification,and the error rate is lower 23.34% than that.