Urban Land Use Classification Based On Sentinel-2A Remote Sensing Images And POI Data

Urban land use is a reflection of the basic state of urban pattern and the regional differences of urban internal functional areas.The study area of this paper is the five administrative districts within the Fourth Ring Road of Changchun City,with a total area of 468.2km~2,and the urban land use types in the study area are divided into 11types.Based on Sentinel-2A satellite remote sensing images,OSM road network data and POI data,random forest algorithm and CART decision tree algorithm were used to classify urban land use in Changchun and explore the advantages and disadvantages of the two classification algorithms in the classification of urban land use.The results show that in the classification results of random forest algorithm,the classification accuracy of the land use types of industrial and mining storage land,education land,sports and cultural facilities land,park and green land is about 80%,and the classification accuracy of business office land and commercial land is better.The classification accuracy of residential land,transportation land,organization land and medical and health land is about 60%.The overall classification accuracy was71%,and the Kappa coefficient was 0.68.In the classification results of CART decision tree algorithm,the classification accuracy of residential land,transportation land and medical and health land is about 50%,the overall classification accuracy is67%,and the Kappa coefficient is 0.63.The classification result of random forest algorithm is more accurate and accurate than that of CART decision tree algorithm in the urban land use classification of Changchun.In the classification of urban land use in Changchun,the higher the mixed degree of land use type,the worse the classification accuracy.Correct selection of sample points in training samples can effectively improve the classification accuracy.In the more detailed classification system,the classification results of urban land use in Changchun have a good classification accuracy,which can provide an effective basis for the urban development of Changchun.