An adaptive approach for multi-media steganography using improved chaotic map and discrete cosine transform

Steganography is the process of hiding secret data within the image, audio or the text files. Video steganography is a technique where the secret data is hidden within video files which allows for covert communication without any suspicious actions. The existing approaches in steganography face issues related to poor retrieval rate due to poor encryption and decryption. So, this research introduces a hybrid approach which is the combination of 2D-improved Henon map and 3D-logistic map. The proposed method encrypts the secret data by utilizing the hybrid improved chaotic map and contains a de-embedding process for retrieving the secret data at the receiver’s end. Also, the cover video frames are chosen using the AlexNet model depending on the score value of frames. The discrete cosine transform (DCT) block wise embedding is used to embed the secret data in the selected frames. The evaluation involves different cover and secret data pairs including video-image, video-audio and image-image. The PSNR for cover-secret data pairs of video-image and video-audio are measured at 47.05 dB and 41.28 dB. Furthermore, the efficiency of hybrid approach is evaluated against the existing techniques based on image-image steganography. In image-image steganography, the PSNR of color pepper image for Hybrid approach is 51.05 dB, which is higher than the existing Permutation Substitution and Boolean Operation. The proposed method shows an MSE of 2.05e \(^{1.2}\) $_{}^{1.2}$ PSNR of 28.4 dB, SSIM of 78.6, and entropy of 2.5 when the image is processed with salt and pepper noise.