JPEG domain image steganography method and system based on generative adversarial network

Abstract

The invention discloses a JPEG domain image steganography method based on a generative adversarial network. The method comprises: generating a tampering probability matrix corresponding to a carrier image DCT coefficient matrix through a generation network; using an analog coding embedding module and a transmittable gradient JPEG transformation module for generating a corresponding secret carryingimage according to the tampering probability matrix; distinguishing the carrier image and the secret-carrying image through the discrimination network, using the classification error as a loss function to perform adversarial training on the generation network and the discrimination network, and finally obtaining the generation network model capable of generating the adaptive steganography cost value. Through combination of the model and a traditional information coding module, secret information is embedded into a carrier image to obtain a secret-carrying image. Compared with a traditional JPEG domain image steganography method, the method has the advantages of being simple in design, easy to implement, high in detection resistance and the like. The invention further discloses a JPEG domain image steganography system. The system comprises a generative adversarial network module, an information coding module and a JPEG transformation module which are obtained through the JPEG domain image steganography method based on the generative adversarial network.

Description

technical field [0001] The present invention relates to the field of image steganography, and more specifically, relates to a method for image steganography in JPEG domain based on generative confrontation network. Background technique [0002] Information hiding is to embed the secret information that needs to be encrypted into the carrier file through some algorithms to generate a secret file, and transmit it through an open channel. The receiver extracts the secret information in the secret file through the corresponding extraction method, and the attack Party cannot obtain secret information from classified documents. Image information hiding is mainly to replace or modify some data of the image, such as pixel values, etc., and to embed secret information into the confidential image without affecting the visual effect of the image. redundancy. Because digital images have the characteristics of large information capacity, easy tampering, and a wide variety of quantities...

AI Technical Summary

Problems solved by technology

The traditional method realizes the adaptiveness of information embedding by setting an additive distortion function, but this method is highly dependent on the designer's experience and prior knowledge, and cannot make timely strategic adjustments according to changes in the steganalysis network, so its Security still has some room for improvement

At present, there is no automatic learning algorithm for the steganographic cost function that adjusts its own strategy in time according to changes in the steganalysis network in the JPEG domain.

Images