Land cover classification method and land cover classification system based on crowdsourced geographic data spatial clustering

Abstract

The invention discloses a land cover classification method and a land cover classification system based on crowdsourced geographic data spatial clustering. The land cover classification method comprises the following steps: acquiring crowdsourced geographic data, and using the crowdsourced geographic data as land cover classification data; carrying out spatial clustering on data points by using coordinate information for expressing spatial positions in the data points of the acquired crowdsourced geographic data, and clustering the data points into a plurality of groups of data points; obtaining the plurality of groups of data points by using the spatial clustering, and delimiting land cover areas containing all the groups of data points; inputting text information of each land cover area into a probabilistic latent semantic analysis model, outputting a land cover theme and a theme weight which consist of terms in the text information of each land cover area by the probabilistic latent semantic analysis model, screening the land cover type corresponding to a theme with highest weight in the text information of the land cover area as a basis for judging the land cover type, and judging the land cover type of a to-be-detected land cover area according to the basis for judging the land cover type.

Description

technical field [0001] The invention belongs to the technical field of land cover classification, and relates to a land cover classification method and system based on spatial clustering of crowd-source geographic data. Background technique [0002] Land cover is an indispensable basic information and key parameter for environmental change research, geographical situation monitoring, and sustainable development planning. It is very important for the study of the energy balance of the earth system, carbon cycle and other biogeochemical cycles, and climate change. meaning. At present, land cover classification mainly uses spectral and texture information in aerospace and aerial remote sensing images, combined with prior knowledge and experience, to extract the type distribution and change information of land cover. This method is time-consuming, labor-intensive, and has a long production cycle, which is difficult to meet the needs of rapid mapping of land cover products. In ...

AI Technical Summary

Problems solved by technology

"Hou, Dongyang, et al."Active collection of landcover sample data from geo-tagged web texts."Remote Sensing 7.5(2015):5805-5827." However, it is difficult to accurately define the type of land cover with text information

"Sitthi, Asamaporn, et al."Exploring Land Use and Land Cover of Geotagged Social-Sensing Images Using Naive Bayes Classifier."Sustainability 8.9(2016):921."Construct land cover classification by extracting land cover information from Flickr images The model realizes land cover classification based on crowdsourced image data. This method requires manual interpretation of a large number of sample images, which is time-consuming and laborious. "Johnson, Brian A., et al."Employing crowdsourced geographic data and multi-temporal / multi-sensor Satellite imagery to monitor land cover change: A case study in an urbanizing region of the Philippines. "Computers, Environment and Urban Systems 64(2017): 184-193." proposed a method to combine text information in crowdsource geographic data with remote sensing images Combined land cover classification method, this method successfully obtained the land use classification map, but it is also limited by the acquisition technology of remote sensing images

[0004] Most of the above methods focus on the image and text features of crowdsourced geographic data, ignoring the influence of the spatial distribution characteristics of the data itself on land cover classification

Images