Micro Landform Classification Method Of Grid DEM Based On Artificial Intelligence

The landform is a component of the natural geographical environment,which is closely related to human survival and life.Landform classification is an important research content of digital terrain analysis,and the digital elevation model has effectively realized the digital expression of surface morphology,especially grid digital elevation model data has particularity and superiority in extracting landform features and is often used as an important data source of digital terrain analysis.With the continuous improvement of grid DEM accuracy and the deepening of landform factor mining by digital terrain analysis scholars,the landform classification scale based on grid digital elevation model has gradually advanced from macro to micro-level,and classification refinement and method automation have become an inevitable trend.Traditional micro landform classification methods based on regular knowledge have some problems,such as a low degree of automation and incomplete classification results.The emergence of Artificial Intelligence and the rapid development of its algorithms provide a new research method for micro landform automatic classification.In this paper,based on digital terrain analysis theory and artificial intelligence as technical support,an AI-based automatic classification method for micro landform based on grid digital elevation model is established,and the classifier models of Support Vector Machine,BP neural network,and Convolutional Neural Network are created respectively,and the whole process automatic classification of the three classifier models is realized by programming.The main research work and achievements are as follows:(1)The development and application of digital elevation model and digital terrain analysis,extraction,and mining of landform factors are elaborated in detail,and the system,basis,and methods of classification of micro landform are summarized.The decision-making scheme of detailed classification of mountain parts and its classification method based on regularized knowledge are emphatically studied.It is pointed out that the essence of the problems existing in traditional micro landform classification,such as low automation and incomplete classification results,is that using combined landform factors to describe landform categories is uncertain and strongly constrained.Based on this,the machine learning method is introduced to analyze the feasibility and adaptability of using AI to classify micro landforms.The analysis results show that the AI method is effective in improving the incomplete rate of grid digital elevation model micro landform classification results.(2)By analyzing the main features of the support vector machine,the adaptability of its application in automatic classification of micro landform is studied,and the grid digital elevation model micro landform classifier model supported by grid digital elevation model algorithm is established.The automatic classification method of grid digital elevation model micro landform based on support vector machine is established,and the whole process automatic classification is designed and programmed.Taking the classification of hill-position as the research object,the grid digital elevation model is used for experimental verification.The results show that compared with the classification method of superposition analysis,the classification method of grid digital elevation model based on support vector machine not only ensures the integrity of the classification results but also improves the classification accuracy.(3)Using the advantages of the BP neural network in geoscience classification,it is introduced into the micro landform classification of the grid digital elevation model.Through the existing classification decision-making scheme and prior knowledge,the typical samples are determined.After the training,inspection,and optimization of sample data,the automatic classification of hill-position in the experimental area is realized,and the classification results are compared and analyzed.The experimental results show that the BP neural network method of grid digital elevation model micro landform classification has obvious advantages over the existing landform factor superposition analysis method,which can not only avoid the tedious data superposition analysis process in the process but also effectively improve the integrity and misclassification rate of classification results.(4)The adaptability of convolution neural networks in micro landform classification is analyzed.Combining with micro landform data,a convolution neural network model suitable for automatic classification of grid digital elevation model micro landform is constructed,and a convolution neural network method for micro landform classification of grid digital elevation model is proposed and its automatic implementation flow is created.Taking the classification of hill-position as a typical example,the experimental results demonstrate the reliability and applicability of convolution neural network algorithms in micro landform classification.In this paper,support vector machine,BP neural network,and convolution neural network algorithms in the field of AI are combined with digital terrain analysis technology,and an artificial intelligence method for micro landform classification of grid digital elevation model is proposed,which solves the problems of low automation and incomplete classification of micro landform classification method based on regular knowledge,expands the application of artificial intelligence in the field of geographic analysis of grid digital elevation model and provides a way for automatic classification of micro landform.