A Comparative Study Of Land Use Classification In Xining Urban Based On Landsat 8 Remote Sensing Imaging

Land resources play the most critical resource in the development of human society. The ecological environment change is closely associated with the land use change and its operations, which is the main reason of the dramatic shift in surface and the material cycle and energy flow.It has more realistic significance to the extraction of land use information in our country. With the rapid economic development and the intensification of urbanization, people increasingly demand the land. The amount of farmland continues to fall. Meanwhile, we have great population but less appropriation of per-capita quantity of the land resources. And the extraction of land use information, mainly force on the cropland and construction land, forest land water system of land use changes in real time, scientific and high precise supervision makes the development of land resources has a scientific basis.(1)According to the classification methods of ENVI include supervised and unsupervised classification. The Maximum Likelihood and the Isodate are in common use classified based on the systematic analysis of the spectrum features which rarely introduces some objective conditions. There is no need to understand the knowledge of Remote Sensing deeply because of the classification method of Decision—tree,(2)There is no need to understand the knowledge of Remote Sensing deeply because of the classification method of Decision—tree, ML and Isodata to reflect the classification accuracy. Its Kappa coefficients are 0.9456, 0.6276 and 0.4123. Obviously the findings about the method of Decision—tree has some reference for establishing.(3)There is a low accuracy in Isodata to reflect the classification accuracy. Its accuracy coefficients are 74.49%?60.50%?96.57%Obviously the findings about the method of Decision—tree has some reference for establishing. From the visual Angle, maximum likelihood method and IsoData classification in accordance with the actual reference image degree is low, and the decision tree classification and actual reference image alignment is higher, can correctly differentiate class.(4)Although the result of the tree classification method is in a high accuracy, compared with the maximum likelihood method. Kappa coefficient increased by 22.08% and 31.80% respectively; Compared with IsoData classification, the overall accuracy and Kappa coefficient increased by 36.07% and 53.33% respectively.But the comprehensive detailed analysis of different land use classification can be seen after not every type is the classification of the decision tree classification precision, such as the use of the water system precision is low, the classification of the decision tree classification(5)The overall accuracy of the tree classification and Kappa coefficient are higher but also complex. High demand for the computer is necessary but useless unconvenient. However, the ML is also a kind of very good method, practice application.