Система интеллектуальной стеганографии с использованием алгоритма AES и SPK
Аннотация
This work presents techniques of a steganography system based on the Least Significant Bit (LSB) substitution, the secret message is an image (grayscale or color). In this work, Alpha channel is added to the cover image (RGB.jpg, where RGB stands for Red, Green, and Blue) to increase the bit-depth of the cover image which become RGBA.png image. Bit-Plane Slicing (BPS) technique applied to the secret image for compressing and decreasing the size of secret data to embed.
To increase the security, two encryption models were applied namely, Simple Private Key (SPK) method, and Advanced Encryption Standard (AES) method, where SPK represent a simple XOR encryption, while AES represent the current encryption standard used today.
Results and comparative studies revealed the effectiveness of the proposed technique in generating the stego images. The size of the secret message can be the same size of the cover image while the distortion of the stego image is very low and the stego image is closed related to the cover image. Results also show that in the embedding capacity of 100%, the normalized cross-correlation is very close to one and the PSNR is up to 38.199dB for the grayscale secret image and 38.155 dB for color secret image in SPK model, while PSNR is up to 38.408 dB for grayscale image and 38.226 dB for color image in AES model.
In addition, a comparison with many steganography approaches based on Wavelet domain and spatial domain is done; the proposed steganography provides embedding capacity and quality of the stego image with PSNR value higher than the comparable methods. This work is programmed and simulated using the MATLAB language.
To increase the security, two encryption models were applied namely, Simple Private Key (SPK) method, and Advanced Encryption Standard (AES) method, where SPK represent a simple XOR encryption, while AES represent the current encryption standard used today.
Results and comparative studies revealed the effectiveness of the proposed technique in generating the stego images. The size of the secret message can be the same size of the cover image while the distortion of the stego image is very low and the stego image is closed related to the cover image. Results also show that in the embedding capacity of 100%, the normalized cross-correlation is very close to one and the PSNR is up to 38.199dB for the grayscale secret image and 38.155 dB for color secret image in SPK model, while PSNR is up to 38.408 dB for grayscale image and 38.226 dB for color image in AES model.
In addition, a comparison with many steganography approaches based on Wavelet domain and spatial domain is done; the proposed steganography provides embedding capacity and quality of the stego image with PSNR value higher than the comparable methods. This work is programmed and simulated using the MATLAB language.