Blind detection of synthetic image based on JPEG quantization distortion
基于JPEG量化失真的合成图像盲检测

This paper proposed an algorithm to blindly detect synthetic image based on JPEG quantization distortion. Firstly, it estimated the original quantization matrix of the synthetic image in terms of its storage format, JPEG or non-JPEG. Secondly, it recompressed the synthetic image using the estimated original quantization matrix, calculated the quantization distortion of the synthetic image before and after compression. Finally, it judged the size of quantization distortion for different region of synthetic image, to realize automatic detection and location of its tampered regions. Experimental results show that the proposed algorithm is effective in detecting synthetic image stored in both JPEG and non-JPEG.