Landscape is a dynamic phenomenon that almost continuously changes. The overall change of a landscape is the result of complex and interacting natural and spontaneous processes and planned actions by man. However, numerous activities by a large number of individuals are not concerted and contribute to the autonomous evolution of the landscape in a similar way as natural processes do. There is a well-established need to detect land use and ecological change so that appropriate policies for the regional sustainable development can be developed. Landscape change detection is considered to be effectively repeated surveillance and needs especially strict protocols to identify landscape change. This paper developed a series of technical frameworks on landscape detection based on Landsat Thematic Mapper (TM) Data. Through human-machine interactive interpretation, the interpretation precision was 92.00% in 1986 and 89.73% in 2000. Based on the interpretation results of TM images and taking Yulin prefecture as a case study area, the area of main landscape types was summarized respectively in 1986 and 2000. The landscape pattern changes in Yulin could be divided into ten types.