Fusing Two Kinds of Virtual Samples for Small Sample Face Recognition

Figure 5

Several original training samples from the ORL face database.