Blurring Detection in Image Forensics Based on the Posterior Probability
Qi Zhang, Haoxin Wang, Jing Hu · International Conference on System science, Engineering design and Manufacturing informatization · 2013
Digital image passive blind forensic techniques aim to examine the authenticity and sources of digital images without relying on any pre-extraction or pre-embedded information. While the image is tampered, in order to eliminate the visual edge distortion caused by splicing, some post-processing operations are usually employed to eliminate the tampering traces. Among them, blurring operation is one of the most commonly used approaches. In this paper, a novel method which can detect manual blurring in the tampered image is proposed. Firstly, a model of the high correlation of pixels in an artificially blurred image is proposed. Then EM algorithm is adopted to estimate the posterior probability that a pixel belong to this model. Finally, the value of the posteriori probability is used for detecting the trace of blurring operation. Experimental results show that this method can effectively detect manual blurring in a tampered image and also has a good robustness against different blurring type, lossy JPEG compression, and global scale operation as well.