Urban change detection is an important field of remote sensing applications. This study presents a method for detecting urban change from remotely sensed data. We proposed the differential image anomaly detection method to improve the moving window method using the Landsat TM data. Changsha urban area was chosen as an example to illustrate the method. The results show that the differential image anomaly de-tection method has higher detection accuracy~ 157 out of 178(88.2%) randomly selected pixels were correct-ly detected. In Changsha urban area, 116 101(5.41%) pixels which was natural landscape turned into urban landscape during 2004-2009. Economic development of industrial areas became the main driving force of the urban change in the study area.