Steganalysis method based on local learning

Abstract

The invention discloses a steganalysis method based on local learning, comprising: constructing a training sample database including positive and negative samples; specific to any sample to be detected, searching for most similar K positive and negative sample couples in the sample database to form a local training set; performing classifier training learning based on the local training set, and adding the constraint of the positive and negative sample couples in a learning process to obtain an optimum classifier by using an optimization algorithm; and discriminating and classifying samples to be detected by the clarifier to obtain detection results of whether the samples to be detected performs steganography. The steganalysis method takes full advantage of local learning which has the advantages of better overcoming the problem of great intra-class change, reducing noise influence and using less priori knowledge, improves steganalysis results, and can be applied to a steganalysis algorithm analysis system based on pattern recognition.

Description

technical field [0001] The invention relates to the field of image and video passive blind evidence collection, in particular to an information hiding detection method based on local learning for image and video information hiding detection. Background technique [0002] Information hiding uses the sensory redundancy of human sensory organs to digital signals to hide a set of secret information (authorization serial number, message or copyright information, etc.) into the carrier information, without affecting the sensory effect and use value of the host signal , making it difficult for possible attackers to judge whether the secret information exists, and even more difficult to intercept, thus ensuring the security of information transmission. Information hiding has become one of the basic communication methods to safely and reliably transmit political, military, economic and other information in the network environment. At present, there are sufficient evidences to prove ...

AI Technical Summary

Problems solved by technology

When using these classifiers, existing methods usually learn a global classification model on all training samples. The problem is that information hiding detection faces the difficulty of large intra-class scatter and small inter-class differences. The learned global classification model often Due to the low generalization performance due to the high complexity, many existing methods can only achieve good detection results under experimental conditions

In addition, the existing global learning methods usually need to train classifiers for each information hiding algorithm and each embedding rate separately. When detecting, it is necessary to know the embedding algorithm and embedding rate used by the sample to be tested in advance, which is very low in practicability.

Images