A Steganalysis Blind Detection Method for JPEG Images

Abstract

The invention discloses a steganalysis blind detection method for JPEG images. Aiming at the problem of modifying the DCT coefficients in the process of steganography for JPEG images, this method combines the widely used adjacent joint density feature extraction algorithm and the bilateral large-distance hypersphere classifier to train the general detection model. In this way, the secret-carrying image generated by the unknown steganographic algorithm is detected. The advantage of the present invention is that most of the current general-purpose blind detection models use a single-class classifier for training, and the detection rate is low, and it is difficult to detect an unknown algorithm by using a two-class classifier for training. However, this method uses two The hypersphere-like classifier can detect unknown algorithms more accurately, and has a higher detection rate than the single-class classifier.

Description

technical field [0001] The invention relates to the technical field of computer information hiding, in particular to a steganalysis blind detection method and a method for establishing a general detection model. Background technique [0002] With the rapid development of network technology, communication technology and multimedia signal processing technology, information hiding, as a new cryptographic technology, has become a new research hotspot in the field of information security. Steganography is an important branch of information hiding technology, which mainly studies how to hide information in public multimedia data to realize covert communication. The corresponding steganalysis studies the attack on steganography, that is, how to detect, extract or destroy hidden secret information. [0003] In response to the development and application requirements of information hiding technology, many steganographic algorithms based on JPEG images, such as F5, MB2, MME, etc., ha...

AI Technical Summary

Problems solved by technology

And this is because the number of classified images we can obtain is limited, and the number of non-confidential images is large. The model trained by data imbalance will have a certain bias, and there will be a relatively large deviation in the accuracy of detection, which will result in relatively high The missed detection rate

There are also single-class models trained with single-class classifiers. Although this method can relatively effectively detect images generated by unknown steganographic algorithms, the relative detection rate of this method is low, and it cannot be used in many cases. Require

Images