Study On The Authenticity Identificaion Techniques In The Digital Photo

Digital photo is one of information carriers and people record some messages with camera at a time or in a moment by these photos. However, its authenticity aroused great concern of scholars and research teams form home and abroad. Image tampering identification techniques provide a strong theoretical basis for identifying the authenticity of digital photos, and it is divided, into two categories:active and passive tampering detection. With the development of image processing technologies, the identification of digital photos authenticity will face to tremendous challenge.This paper represents the development history and the significance of digital image tampering identification techniques, analyzes and summarizes the research findings. According to this, this paper’s focus is to research the identification of the digital photo authenticity. Generally, doctor detection techniques are divided into two classes. The one is the active tampering identification, and we provides a new schedule to detect the image regular clipping operation based on scale invariant feature transform (SIFT). The feature points in the photographs are extracted by the algorithm of SIFT, are matched by using the method of nearest neighbor, and then the matching keys are taken. According to the invariant relevant relation of location and space geometry, clipping level, scale rate and the location of cutting area can be calculated from the testing photos.The other is the passive tamper authentication, and this paper presents the doctor authentication method based on the consistency of the natural properties in the picture. Two detection methods based on separately the lighting direction and the shadows are combined to wide the detection range for the tampered photos. Lambertian model can get the illumination direction from the object surface, and the direction area of lighting source can be estimated with cast shadows of objects in the picture. If the lighting direction is not located in the direction area, the digital photo is doctored. This method not only fits the situation that there are multiple objects in the picture, but also the scene that there are only one object. With regard to uniqueness of the illumination chromaticity in the picture, the estimated value of illumination chromaticity is quantified and analyzed, and used in the tampering detection of digital photos. Based on the theories of dichromatic reflectance mode, inverse-intensity chromaticity (IIC) space and Hough space, the chromaticity values can be estimated about three color channels. Next, the difference model of the chromaticity between different objects is built, and finally the difference value is compared with threshold value. If the difference of the illumination chromaticity exceeds the threshold, the consistency of illumination chromaticity should be broken. Hence, the image can be authenticated as doctored by this way.