Application Research And Intelligent Extraction Of Vegetation From Domestic High-resolution Remote Sensing Image

As an important part of the terrestrial ecosystem,vegetation plays an indispensable role in water and soil conservation,biological carbon sequestration and climate regulation.With the improvement of the resolution of remote sensing images,the high-precision extraction of vegetation information is of great significance for understanding the changing laws of surface vegetation and evaluating ecological regions.The normalized difference vegetation index method is the most commonly used greenness vegetation index method in vegetation extraction,but it is susceptible to the influence of vegetation activity,and it is difficult to reflect the withered vegetation information.However,the existing deep learning research mainly focuses on the extraction of green vegetation,and has not discussed the extraction of withered vegetation in depth.Aiming at the problem that the existing vegetation extraction methods are difficult to extract the yellow vegetation information,and it is difficult to realize the vegetation cross-season extraction.This paper proposes a deep learning semantic segmentation network of vegetation extraction method based on the feature separation mechanism based on the Gao Fen-2 satellite data.This method can scruples of vegetation in the complex background of high-level semantic information,can effective use of vegetation characteristic information,to solve the current vegetation extraction method is difficult to effectively reflects the problems of the yellow vegetation information,realize the complete extracting vegetation across a season.And use the method of this paper to extract and analyze the changes of vegetation in Luyang City,Hefei City,Anhui Province,and realize the intelligent monitoring of vegetation and the comprehensive evaluation of the distribution of park green space.The results of this paper can provide data reference for urban ecological environment evaluation and vegetation application research.The main result of this paper are as follows:1.In view of the problems that existing vegetation extraction methods are difficult to reflect the information of withered vegetation,difficult to realize the problem of vegetation cross-season extraction,a cross-season vegetation extraction model is constructed.Based on Dense Net model framework,the network adds a feature separation mechanism that combines separable convolution and spatial pyramid on the basis of Dense Net.The atrous spatial pyramid effectively reduces the loss of information while acquiring spatial features of different scales.This network takes the high-level semantic information of vegetation into account in complex background.The feature information is enhanced,and the accuracy of the model is improved.In order to reduce the calculation amount and the parameter amount of the atrous spatial pyramid,a separable convolution layer is used to replace its original convolution layer.This method solves the current vegetation extraction method is difficult to effectively reflects the problems of the yellow vegetation information,realize the complete extracting vegetation across a season.2.The effectiveness of the proposed method is verified compared with traditional vegetation extraction methods and common deep learning methods.In order to verify the effectiveness and accuracy of the vegetation extraction method in this paper,this paper conducts comparative analysis with traditional vegetation extraction methods(NDVI method,maximum likelihood method,object-oriented method)and other deep learning methods(Deeplab-V3+,U-Net,Dense Net)under the same data set.Through comparative analysis of extraction results,it is found that the proposed method is superior to traditional vegetation extraction methods and other deep learning methods in many evaluation indexes,the method presented in this paper has great advantages over traditional vegetation extraction methods and other deep learning methods in the complete extraction of cross-seasonal vegetation.And the F1 score reaches91.91%,the overall accuracy reaches 92.79%,the intersection ratio reaches 85.10%.The generalization of vegetation extraction is verified on the remote sensing images of GF-1 and GF-6.The results show that the method proposed in this paper has a certain general ability.It can realize the automatic and high-precision extraction of vegetation from high resolution remote sensing images.3.Intelligent monitoring and application of vegetation have been realized.The method in this paper is used to realize the intelligent extraction of multi-time series remote sensing images of the vegetation in Luyang urban area of Hefei City,Anhui Province,calculate the vegetation area of Luyang urban area in 2015,2017 and 2019,and analyze the influencing factors of the change.At the same time,based on the extracted vegetation information and combined with Google image,park areas in Luyang urban area are screened out.The network analysis method is used to simulate the park green space and residents' travel conditions in the study area,and the spatial distribution and service area of the park in Luyang urban area are comprehensively evaluated from the perspective of time distance.