The Compaq database [4] consists of approximately 4000 skin labeled images and 8000 non-skin images. A typical image contains something in the order of 10,000 pixels. The database is split into two halves: one half is used for training, the other for ’un-seen’ testing. The training set uses the first 2000 skin images and the first 4000 non-skin images. These 6000 images are used to generate the SPM. The first 2000 skin images are then used to establish thresholds for 95% skin acceptance for all approaches.