探索城市改造过程中景观的变化规律对城市景观变化的预测有重要的指导意义。以沈阳市铁西区1997、2002、2005、2008年建筑物三维信息为基础数据。按照建筑物使用性质将其分为居住、工业、商业、教育与医疗、其他建筑5类,按照建筑物的高度将其分为平房、低层、多层、中高层、高层和超高层6类。采用转移矩阵法分析了不同时段建筑类型的更替规律。并将探索性空间数据分析法应用于建筑改造空间分布特征的研究。结果显示: 1997到2002年,居住建筑是建筑改造的主体。2002年后,工业建筑成为建筑改造的主体。1997到2008年,不同高度类型的建筑更替顺序为:平房→低层建筑→多层建筑。新增建筑先以低层为主,后以中高层,逐渐以高层为主。建筑改造在空间上并不是随机分布的,而是存在显著的空间正相关关系。在局部空间范围内,建筑改造呈现成片集中改造的现象。;To explore the variation law of architectural landscape is important for forecasting the variation in urban landscape in the process of urban renewal. It was mainly studied the evolution law of architectural landscape and spatial distribution features of building transformation in the Tiexi District, Shenyang, China, during the urban renewal process in this paper.<br> In this article, the three-dimensional information of buildings in the Tiexi District was extracted from aerial photos (1997) and QuickBird imageries (2002, 2005 and 2008) based on ArcGIS, Erdas and Barista software. We divided the buildings into five types as Residential building, Industrial building, Commercial building, Educational and Medical building, and Others according to the usage properties, and six types as Bungalow, Low-rise building, Muti-story building, Mid-tall story building, High-rise building, and Super-high building according to the height, combining our research purpose with the principle of landscape classification. The conversion-in contribution rate, conversion-out contribution rate and retention ratio were used as the indexes of the transfer matrix method analysis to detect the change characteristics of architectural landscape types and their position and function. We divided the research area into 457 blocks of 0.3km×0.3km, and caculated the transformation area ratio of buildings in each block. Both the global spatial autocorrelation Moran's I indicating overall clustering that exists in a dataset and the local spatial autocorrelation (LISA) cluster map dentifying ‘interesting’ locations and assessing the extent of spatial heterogeneity of the exploratory spatial data analysis were used to study the spacial distribution pattern of building transformation.<br> The results showed that the main transformation of architecture landscape types was between the Not buildings and buildings (Residential building, Industrial building and Commercial building; Bungalow, Low-rise building and Multi-story building) from 1997 to 2008. There was significant difference existed in the transformation of building types and the spatial distribution in different time intervals.Residential buildings were the leading actor of architecture transformation from 1997 to 2002, while Industrial buildings were after 2002. The buildings were replaced firstly the Bungalow, secondly the Low-rise building, then the Muti-story building, and the newly increased buildings were firstly the Low-rise building, secondly the Mid-tall story building, then the High-rise building from 1997 to 2008. The Moran's I of global spatial autocorrelation indicated that the building transformation was not distributing randomly but with prominent positive relation in space. The cluster map of local spatial autocorrelation presented concentrated building transformation in the local spatial scope. The position of the centralized transformation changed constantly over time. The implementation of urban renewal policies and transformation of regional function were the main factors for variation in architecture landscape pattern in the Tiexi District from 1997 to 2008.