The emergence of digital image and video acquisition devices has today lead to an ever growing amount of images taken and video footage produced, asking for methods and techniques to search and analyze this media for particular content. Unfortunately, advances in digital media acquisition are not only employed to record common settings like vacations or birthday parties. Instead, they are also utilized to produce illegal pornographic media by documenting Child Sexual Abuse (CSA). With the various ways of media sharing offered via the Internet, such as P2P networks or file sharing platforms, a wide distribution of digital media is possible. Both aspects, simple media creation and its distribution, confront law enforcement units with constantly increasing numbers of media files to be analyzed when pursuing cases of CSA or possession of such media. Today, the amounts of data have grown beyond the capacities of human and investigation resources.

This thesis provides approaches for scene and object retrieval, as well as concept detection to create methods and systems to support law enforcement agencies in their fight against sexual child abuse.