ABSTRACT
The study on the source forensics of image is intensive at home and abroad, and has proposed many effective methods, but most of them are on digital cameras, only a few are on mobile camera. Besides, most methods are based on characteristics, while few on non-characteristics. Cao Hong et al. proposed a method of using demosaicking through three kinds of images extracted by the interpolation principle of CFA to realize mobile image origin identification [1]. Alles et al. continued to use the origin identification of digital camera. They proposed a method which is based on sensor pattern noise (PN) to determine the source of heavily compressed images from low-resolution cameras [2]. One of the important features of PRNU is its high-frequency components, which have nothing to do with the scene, but it is comparatively stable among the life of the camera. Fridrich et al. first proposed that we could get the residual images by low-pass filter, and then stack these images for average to obtain PRNU [3]. But this way is easy to be influenced by scene noise and interpolation noise of CFA. Therefore, the papers [4] and [5] proposed a method to enhance the useful components of the residual noise and to suppress the amplitude of scene noise and other noises by the observation of ideal camera’s fingerprint to detect the source of images. The paper [6] proposed a method of using large amplitude in pattern noise to restrain scenic noisy amplitude. Combined with CFA filtering process in camera, paper [7] proposed a strategy of extracting noise based on the pattern noise of DPRNU. Chan-Tsun Li proposed a method of using CD-PRNU
to detect the source of cell phone camera image [8]. In summary, CFA interpolation and pattern noise of image can be used to detect the source of the images, so this paper combines the advantages of both to detect the image source of mobile phone.
