||When a large-scale disaster happens, family separated without any preparation. In shelters, we need to provide the safety information about the families. With the safety information, we decrease the useless and the meaningless activities of people. This increases the QOL in the shelters. In the case, people have their family images. This system using face recognition function. The function is based on PCA (Principal Component Analysis), and PCA algorithm can get precision features. With these features, face recognition function to execute. In the case, much peoples cannot access any face images of their families. The system using face clustering function of the AAM (Active Appearance Model). We use AAM to find the geometric feature. Counting the distance between features, and face clustering by these distance. In the cluster, compared with the center point, nearest face image became center face image.