Using aerial television images taken after the 1995 Hyogoken-Nanbu earthquake, automated detection of building damage based on the maximum likelihood classifier was attempted. Each class for the classification was defined by four attributes of either including color information or only edge texture. In both cases, the estimated results were in good agreement with the actual distribution of severely damaged buildings and the results of previous studies based on the multi-level slice method. In particular, the method based on edge information can be applied not only to adjacent images but also to other images taken after other recent earthquakes without selecting training data. Finally, application of the automated damage detection methods to early damage assessment was discussed.