Research on a Feature-Based JPEG General Detection Algorithm

JPEG format has become the most commonly used compressed format in the present digital image transmission and storage for its excellent compression performance, better image quality and flexible compression selection. This paper studies a JPEG general detector, analyzes various kinds of eigenvectors which are extracted from the image to be detected and used for classifier learning and detection and further proposes a method of calculating statistical properties by regional division, which optimizes the method of extracting eigenvectors and is a research direction of improving the performances of general detectors in the future.