Extensive dual tree complex wavelet transform-based image steganography using SVD and CNN subspace

Abstract

The secret data is hidden behind a cover image using digital image steganography technology that prevents the human visual system (HVS) from identifying it. The proposed image steganography method involves scrambling the secret image and then embedding it within the cover image using the dual-tree complex wavelet transform (DTCWT) and singular value decomposition (SVD) technique. The stego image is generated by applying the inverse of DTCWT on the processed embedded image. The PSNR values for the stego and extracted secret images are 45.51 to 30.08 dB and 45.93 dB, respectively, with an SSIM greater than 0.99 in all cases. In the final step of the proposed method, a convolutional neural network (CNN) is utilized to enhance the quality of the extracted secret image. The effectiveness of the proposed method is confirmed by the comparative results.