Research On The Method Of Extracting Karst Vegetation From High Resolution Remote Sensing Image

The Karst area in southwest China is about 513,600 square kilometers,accounting for 5.35% of the total land area.Forest,shrubs and other natural vegetation grow on the Karst peaks,which plays a crucial role in maintaining the ecological balance of Karst areas.Obtaining coverage information of Karst vegetation and monitoring the change of Karst vegetation is an important task of forestry forest survey.However,the investigation of artificial forests in Karst vegetation in a large area consumes a lot of manpower,material resources,financial resources and time,and the results of the survey cannot reflect the specific conditions of large-area Karst vegetation in time.The development of remote sensing technology,especially the emergence of high-resolution remote sensing images,provides the only economic,rapid and feasible applicability method for the detailed investigation of large-area Karst vegetation information.However,there is obvious spectral overlap between Karst vegetation and ground vegetation such as adjacent crops in high-resolution remote sensing images.The appearance of "the same spectrum of foreign matters" introduces certain errors to the automatic remote sensing extraction of Karst vegetation,which is easy to cause "false identification and missed identification",and it is difficult to ensure the accuracy of Karst vegetation information extraction.Therefore,it is of great significance to study the method of extracting karst vegetation in high-resolution remote sensing images quickly and accurately,and to develop software to process remote sensing big data and monitor the vegetation cover change of large-area Karst peaks.In this paper,the forest vegetation in the karst mountain peak is extracted as the final purpose.Some of the landforms taken by satellites in the suburbs of Guilin are used as research areas,and high-resolution remote sensing images are used to extract forest vegetation.The method of forest vegetation extraction based on SVM and HSI is expounded.The calculation method of color information in high-resolution images and the specific operation steps of support vector machine are introduced.The experimental results are compared and analyzed.Based on Visual Studio,a remote sensing image forest vegetation extraction system is designed to facilitate the comparison of various methods,and finally summarizes and prospects the work of the paper research.The main tasks of this paper are:(1)In order to achieve the purpose of automatic extraction of karst vegetation information in high-resolution remote sensing images,high-resolution remote sensing images are performed using classical algorithms such as Otsu method and maximum entropy method,and new methods combining FRFCM,convolutional neural network deep learning algorithm.The extraction of forest vegetation has given the new method a certain advantage in extraction accuracy.(2)In order to reduce the interference of "the same spectrum of foreign matters" phenomenon and improve the accuracy of karst vegetation extraction,a remote sensing image segmentation method based on SVM and HSI spatial features of support vector machine(SVM)is proposed for the problem of general methods.The color information is trained,classified and identified,and the classified image is removed from the non-green background operation,and finally converted into the HSI space for morphological opening and closing reconstruction to obtain the final extraction target.Through comparison experiments,it can be found that the extraction accuracy and operation speed of the method are obviously superior to other methods,and meet the requirements of the forestry department to process remote sensing big data quickly and accurately.(3)Using Visual Studio programming to realize the "Automatic Extraction System of Karst Forest Vegetation",designing a visual interface and facilitating operation.It can compare the forest vegetation extraction results and running processing time under different algorithms,realize rapid and accurate monitoring of remote forest big data karst forest vegetation changes.