Object-oriented Land Use Classification Based on Landsat Images: A Case Study of the Lower Liaohe Plain

Based on Landsat OLI images, support vector machines and random forest classification methods are used to object-oriented classify land use in the lower Liaohe Plain, and the influence of DEM factor on classification accuracy is also discussed.The results show that: 1)The overall accuracy and Kappa coefficient of the random forest classification in the study area are better than the results of the support vector machine classification.At the same time, referring to the DEM factor helps to improve the classification accuracy.The overall accuracy is up to 91.80%.2)The classification accuracy of paddy field and construction land under the support vector machine classification method in the study area is higher; while dry land, woodland, water and unused land are more suitable for random forest classification.DEM factors can improve the classification accuracy of paddy field, dry land, water and unused land, while reducing the classification accuracy of woodland to a certain extent.3)In 2015, the land use in the study area was mainly cultivated land, followed by construction land and woodland.The area of water, unused land and grassland was relatively small.The object-oriented land use classification of medium-resolution remote sensing images has good applicability and can provide a reference for regional land use information extraction. © 2021, The Editorial Board of Journal of Basic Science and Engineering. All right reserved.