An icon (from Greek εικων eikon "image") is a religious work of art, most commonly a painting, from Eastern Christianity and in certain Eastern Catholic churches. For studying the culture behind the icons, it is important to know the similarities and differences between them. This paper propose a method of clustering west Icons using facial shape features. The facial shape is represented by the landmarks of faces. A k-means method is employed for the clustering. The experimental results show that the proposed method can effectively cluster the Icons.