Abstract: The Laoshan scenic area is an ecological barrier and biodiversity conservation area in Nanjing City. The dominant ecosystem in Laoshan is forest, its ecological position is extremely important. This study aims to analyze the dynamic changes of the landscape pattern in the scenic area for the past 17 years. Local land use information based on Landsat images from 2001, 2005, 2009, 2013 and 2017 was extracted, using the technical method of object-oriented fuzzy classification, C5.0 decision tree classification, pixel-based maximum likelihood, artificial neural network and support vector machine. In addition, Fragstats 4.2 software was used to extract multiple landscape pattern indexes, for the comparison of dynamic landscape pattern from 2001 to 2017. Results show that the object-oriented fuzzy classification method had the highest overall accuracy (91.79%) and Kappa coefficient was 0.83. With the continuous construction and reform, the degree of landscape fragmentation in Laoshan scenic area has deepened. From 2001 to 2017, the number of patches (NP) showed an increasing trend, and reached 653 in 2017, with the patch density of 5.928 1 blocks·km^-2. In 2017, contagion index (CONTAG) increased compared with that of 2001, reaching 59.41%, and the connectivity of dominant patch types has improved. The Shannon's diversity index (SHDI) value fluctuated in a narrow range, reaching a maximum of 0.725 5 in 2009. With the development of construction to a mature state, the overall development process slowly enters into an orderly state, and scenic area in Laoshan has gradually developed towards harmony and stability.

Key words: forest resource, land type information extraction, object-oriented, landscape pattern, Landsat