Research And Implementation Of A Recognition Method Of Terrain Features Based On Statistical Distribution

Terrain features are the basis for reflecting terrain and landform and describing the terrain structure.Valleys and ridges,as important terrain features,are of great significance for the study and analysis of terrain systematically.How to quickly and accurately identify and extract valley and ridge lines has always been a hot issue in the field of terrain analysis.This thesis mainly studies how to use the statistical analysis method in modern geography to quantitatively analyze terrain features,and realizes the recognition of terrain features based on the statistical distribution of terrain data.The main work done in this thesis is summarized as follows.Firstly,this thesis analyzes the various existing algorithms of terrain feature recognition.The basic knowledge of topography,the theory of topographic evolution,the theory of mountain system and other important basic theories,as well as the mathematical methods and statistical analysis methods of data in topography are summarized.And then,the trend surface analysis method and the Gaussian mixture model which are closely related to the research contents of this thesis are mainly studied.Secondly,this thesis collates and filters the terrain sample data,analyses the main characteristics of sample data,deeply studies the trend surface analysis method and makes adaptive improvement,at the same time,improves the polynomial function model and the Gaussian mixture model.And then,this thesis has carried out several sample experiments,quantitatively evaluates the experimental results,identifies the optimal function model,and summarizes the statistical distribution of terrain data.Thirdly,in view of the complexity of geographic system,this thesis proposes a dimension reduction decomposition scheme,and improves the terrain layered model,which reduces the complexity of data processing and improves the computational efficiency.And then this thesis determines the process of solving the terrain feature parameters and formulates the rules for determining the terrain data.On this basis,a terrain feature recognition method based on statistical distribution is proposed.At last,the method proposed in this thesis is programmed to achieve,and terrain map experiments have been completed,which validates and evaluates the performance of the method studied.The experimental results show that the terrain feature recognition method studied and designed in this thesis can accurately identify ridges and valleys under various terrain conditions,and extract complete and clear ridge and valley lines.The functional integrity,applicability and recognition performance of this method have reached the expected goal of the study,which has certain reference significance for the related research in the field of terrain feature recognition.