To enhance the security, visual consistency and anti-steganalysis capability of image steganography, the reversible steganographic network DISG-Net is proposed. The model ensures the security of secret information through an innovatively designed QR code-based dual encryption method, and employs 16 wavelet transform-based reversible blocks to achieve high-fidelity embedding and reversible recovery. Meanwhile, a BYOL self-supervised discriminator is introduced to constrain the feature distribution, making the generated results natural and difficult to detect. By incorporating multiple objective losses, including reconstruction, guidance, contrastive, and hashing losses, the framework achieves both visual consistency and precise recovery of secret information. Experimental results demonstrate that DISG-Net outperforms existing methods in terms of image quality and information security, offering a high-fidelity and tamper-resistant information embedding solution for printing and packaging, thereby enhancing anti-counterfeiting and information protection capabilities.