Nowadays, images are commonly found in Facebook, YouTube, WeChat and WWW all over the world due to the prevalence of digital cameras, smart phones, 3G/4G, cloud storage and social networks. Often images are an important source of evidence in law enforcement and investigative cases. However, with modern powerful image editing software, images can be easily modified without leaving visible signs of being altered. This raises the problem of proving the authenticity of images. The problem can be addressed with image forensic techniques, which study the underlying data behind the visual contents. This is based on the fact that image manipulations tend to leave various traces on the images. Although the traces may be imperceptible to the human eyes, they can be detectable with suitable forensic procedure. In this thesis, we propose a camera model identification method exploiting the inherent inter-channel correlation formed during the image acquisition in which a color filter array (CFA) light sensor is used to capture colored light from a scene and demosaicking is performed to generate a full color image. Often image processing steps would disturb the regularity of such correlation, leaving a detectable trace. Our proposed method seeks to detect such trace as a forensic means to detect possible image tampering. On the other hand, when forgers know of possible forensic means that may be employed, they would seek to develop anti-forensic means to hide the traces of their tampering operations. Thus, the forensics researchers need to anticipate possible anti-forensics methods in order to improve their forensic techniques. In this thesis, we introduce a possible anti-forensic method that forgers may use to counter an existing contrast enhancement forgery forensic technique. We will show that the anti-forensic method can be quite effective, which suggests that the current forensic detection for contrast enhancement needs to be improved.