Three-dimensional complex microtopography recognition method and device, computer equipment and storage medium

Abstract

The embodiment of the invention discloses a three-dimensional complex microtopography recognition method and device, computer equipment and a storage medium. The method comprises the following steps: acquiring DEM data to be identified; inputting the to-be-identified DEM data into a classification model to identify a micro-topography category so as to obtain a corresponding micro-topography category; wherein the classification model is obtained by adopting a DEM encoder and a Transformation Group-based data dimension reduction mode to reduce DEM data to serve as a sample set, and training a convolutional neural network in combination with a parameterized microtopography classification system and a complex microtopography automatic pre-clustering classification algorithm. By implementing the method provided by the embodiment of the invention, the problems that the coverage of a classification system is incomplete and inflexible, DEM data is complex in texture and redundant, micro-terrain classification is easy to be wrongly classified, the sample size is unbalanced, manual intervention is needed, and the clustering number is not easy to determine can be solved.

Description

technical field [0001] The present invention relates to a terrain recognition method, more specifically to a three-dimensional complex micro terrain recognition method, device, computer equipment and storage medium. Background technique [0002] A small section of the transmission line, even between one or two base towers, under the influence of systematic weather and special terrain, makes certain climatic factors in this area particularly enhanced, exceeding the ice and wind conditions designed for this area, thus Areas that may endanger the operation of transmission lines are called "micro-topographic micro-meteorological areas". In recent years, many researchers have realized that micro-topography and micro-meteorology have a greater impact on transmission towers, and have tried to analyze them from meteorology, terrain, etc., but few researchers have comprehensively classified and analyzed micro-topography and micro-meteorology areas. [0003] Geomorphology refers to t...

AI Technical Summary

Problems solved by technology

[0004] Micro-terrain classification technology includes the existing micro-terrain classification system, micro-terrain DEM (Digital Elevation Model, Digital Elevation Model) coding and the system of first clustering and then classification; the current micro-terrain classification system used by researchers cannot be used for all possible terrain For coverage, there are only micro-topography that a few researchers think may have an impact on the transmission line. Similar terrain cannot be parameterized and quantified, and it is prone to misclassification. It is necessary to manually search for samples to perform supervised classification, which is very consuming. When; for DEM coding, the Geomorphons algorithm is currently the most cited method, and the micro-terrain classification can be carried out on this basis, but the Geomorphons algorithm only considers the height of each point in the nine-square grid relative to the center point. In the Jiugongge DEM, the pairwise size of the nine points is not considered

This will lose many DEM textures

But if overthinking every possible texture situation of the DEM, the sample size will be huge

For the classification of DEM, most scholars generally use the method of clustering first and then classifying, and manual intervention is required in the middle. For relatively similar categories, misclassification is prone to occur, and in the actual terrain, the number of DEM samples of different categories are different, it is easy to affect the effect of the classification algorithm

[0005] Therefore, it is necessary to design a method to solve the problem of incomplete and inflexible classification system coverage, complex texture and redundancy of DEM data, and easy misclassification of micro-topographic classification, unbalanced sample size, manual intervention, and difficult number of clusters. identified problem

Images