In criminal cases involving child sexual exploitation, critical evidence can be obtained from images and video. Footage is commonly produced during the committing of these crimes, although they do not always show the faces of those involved. A feature that is common though, is the hand. It is possible to analysis these images, and extract identifiable features of the hand which can be used for identification. One such feature is skin pigment. The task of finding and annotating the pigment though is long and laborious one.
Computer vision can be used to make this a semi-automated process. The hand is segmented from the image using user inputted values. Then regions likely to produce the most reliable results are found. Hair is then located and replaced with skin. A LoG detector is used to identify the positions of pigment. With the results displayed to the user. Separate experiments are carried out on classifying the detected features.