Analyzing the accuracy of detecting steganograms formed by adaptive steganographic methods when using artificial neural networks

Abstract

This paper reports a comparative analysis of accuracy in the detection of steganograms formed according to adaptive steganographic methods, using steganography detectors based on common and specialized types of artificial neural networks. The results of the review of modern convolutional neural networks applied for the tasks of digital image stegoanalysis have established that the accuracy of operating the steganography detectors based on these networks is significantly compromised when processing image packets characterized by a significant variability of statistical parameters. The performance accuracy of steganography detectors based on the modern statistical model of container images maxSRMd2 has been investigated, as well as on the latest convolutional and «hybrid» artificial neural networks, in particular, GB-Ras and ASSAF networks, when detecting steganograms formed according to the adaptive steganographic methods HUGO and MiPOD. It was established that the use of the statistical model maxSRMd2 makes it possible to significantly (up to 30 %) improve the accuracy of steganogram detection in the case of analyzing those images that are characterized by a high level of natural noise. It was found that the use of the ASSAF network makes it possible to significantly (up to 35 %) reduce an error of steganogram detection compared to current steganography detectors based on the GB-Ras network and the maxSRMd2 statistical model. It was determined that the high accuracy of the ASSAF network-based steganography detector is maintained even in the most difficult case of image processing with high noise and poor filling of the container image with stegodata (less than 10 %). The results reported here are of theoretical interest for designing high-precision steganography detectors capable of working under conditions of high variability in image parameters.

Destruction of stego images formed by adaptive embedding methods with dictionary learning methods

“…The majority of research in the domain of digital images steganalysis is aimed at development of stegdetectors with extra low error rate [2]. These detectors allows reliably detecting wide range of known embedding methods even for the most difficult cases of low a progonov@gmail.com CI payload (less than 10%) [3]. However, performance of modern stegdetectors considerably depends on prior information about used embedding methods.…”

Performance Analysis Of Stego Image Calibration With Usage Of Denoising Autoencoders

“…The important feature of СNN is ability to adjust their parameters during training to minimize predefined objective function, for example, total variation of image's pixel brightness [25]. Results of performance evaluation of state-of-the-art architectures for CNN proved effectiveness of this approach for improving SD detection accuracy in case of processing widespread embedding methods [26,27]. However, performance of CNN-based image calibration highly depends on either prior information of used embedding methods, or samples of stego images that can be used for CNN tuning [27].…”

“…Results of performance evaluation of state-of-the-art architectures for CNN proved effectiveness of this approach for improving SD detection accuracy in case of processing widespread embedding methods [26,27]. However, performance of CNN-based image calibration highly depends on either prior information of used embedding methods, or samples of stego images that can be used for CNN tuning [27]. This requires often retraining of CNN-based SD to preserve fixed detection accuracy for new set of images that is computationintensive operations even by usage of pre-trained models [10].…”