Gamma Statistic Kakutani Fixed Point and Equilibrium Generative Adversarial Network Based Secure Steganography

Images have been frequently utilized as ideal conditions for hiding information through the employment of steganography algorithms. The information for hiding ranges between document digitization, another secret image where hidden data is said to be embedded without the intervention of humans. Until now, an abundance of steganography algorithms are found in the state-of-the-art works, as well as steganalysis techniques, dedicated to hidden information detection in files.Present-day Steganography algorithms depend on machine learning (ML) and deep learning (DL) for embedding as much as secret text images as probableas reducing visual changes in a given input image (i.e., cover image) with improved accuracy.However, the embedding rate and bit error rate were focused minimally. Following this direction, this article endeavors to exemplify that a Generative Adversarial Network (GAN) can be utilized to enhance the potentiality of spatial domain steganalysis method and to position secret text message information with minimal image alteration (i.e., improving the embedding rate and bit error rate) significantly. In this work, a method called, Gamma Statistic Kakutani Fixed-point and Equilibrium Generative Adversarial Network-based (GSKF-EGAN) secure steganography is proposed. First, a Gamma Statistic Histogram Equalization-based Preprocessing model is designed by fine-tuning the gamma coefficient and equalization for both the cover image and secret text image to circumvent high peak and ensure optimal illumination simultaneously. Second, the Kakutani Fixed-point and Equilibrium Generative Adversarial Network-based Steganography model is designed with the processed images as input. Here by employing the Kakutani Fixed-point ensures a significant embedding rate during the embedding process and using Strategic Equilibrium for validation ensures minimum distortion or bit error rate during the extraction process. Also, experimental results reveal that retrieved secret message data generated by the GSKF-EGAN method with minimum distortion or bit error rate, therefore enhancing retrieved image quality with a higher embedding rate than those generated by conventional image Steganography techniques.