摘要:With the rapid development of social economy, science and technology, the rapid development of Chinese cities has led to the great changes of regional society and economy. Reducing the adverse effects of economic growth and social progress on ecologi-cal environment landscape pattern and promoting the virtuous cycle of natural resource system and social economic system have become the socialgoals. Eco-logical environment landscape image feature extrac-tion can provide timely and effective regional envi-ronment landscape information, which is important to analyze, predict and comprehensively evaluate the dynamic changes of landscape environment caused by the implementation of regional planning. In this paper, the Landsat-TM remote sensing images in 2010 and 2017 are used to extract and analyze the features of the remote sensing images by using the maximum likelihood classification method. The re-sults show that the overall classification accuracy and kappa coefficient of TM images in 2010 are 89.2409% and 0.8532.respectively, and the overall classification accuracy and kappa coefficient of TM images in 2017 are 86.2134%and 0.8012.respec-tively. The deviation rate of classified data in 2010 and2017 is not large. The deviation rate of woodland landscape is the smallest (2.89% for 2010 and 1.28%for 2017). followed by construction land landscape (3.54%for 2010 and 3.13%for 2017),and the devi-ation rate of other land landscape is the largest (7.85% for 2010 and 11.23% for 2017).