Abstract: Blind digital image forensics, a technology for detecting image authenticity and source without relying on any pre-extraction or pre-embedded information, is emerging as a new hotspot with broad prospect in the multimedia security field. First, the main problems and tasks of image forensics are briefly reviewed. Then, according to the forensics characteristics used by image authentication, the techniques of blind image forensics for authenticity detection are divided into three categories: techniques based on the traces left by the process of image forgery, techniques based on the consistency of imaging equipment, and techniques based on the statistical characteristics of natural images. The basic characteristics, typical methods, as well as performance comparison and analysis of various algorithms are summarized in detail for each category. So the latest related works are analyzed, and discussion on unresolved problems and future directions is presented.