Blind Detection Of JPEG Images And IJG OPENCV Static Library Based On

In today’s digital age, it is becoming increasingly easy to modify and edit digital photos with the widespread use of image processing software and editing tools. Fake image for the news media or forensics, it will cause a lot of adverse impact on the society and the general public.Currently, the images from most of the image acquisition device are experienced lossy JPEG image compression, and the images spread on the Internet are mostly in JPEG format images. Therefore, JPEG format image blind detection research has broad application prospects.This thesis is mainly carried out the following work:(1)、According to the existing JPEG image authenticity filtering algorithms, this paper is to start from the statistical properties of the partial numerical frequency of AC coefficients between peak difference, and propose a simple detection method for easy distinction between whether the JPEG image experienced secondary compression. The method does not need the help of the the S VM vector-based method of training can also reach more than97%detection rate.(2)、For secondary compression tampering JPEG image, learning from the existing methods of Blocking Artifact Measure(BAM) and JPEG Quantization Coherence Measurement, in accordance with the JPEG format block consistent coverage and dislocation coverage characteristics, this paper simulates the JPEG secondary compression process and defines L1-norm of the difference between the two images before and after as its measurement. This method can effectively detect image authenticity, determines compression factor for the first time according to the peak point determination and delineates tampered area in the original image.(3)、Based on OPENCV and IJG static library, the author developed a strong operational JPEG image blind detection software, it is proved that the software has some practical value and significance on the test of CASIA image database.