Alireza Shahanaghi - School of Electrical and Computer Engineering, College of Engineering, University of Tehran, Tehran, Iran
Mohammad Ali Ali Akhaee - School of Electrical and Computer Engineering, College of Engineering, University of Tehran, Tehran, Iran
Saeed Sarreshtedari - School of Electrical and Computer Engineering, College of Engineering, University of Tehran, Tehran, Iran
Ramin Toosi - School of Electrical and Computer Engineering, College of Engineering, University of Tehran, Tehran, Iran

LSB matching techniques are widely applied in the field of image steganography. In such algorithms, pixel values of each group must be changed in a way that a predefined function of the pixel group matches the secret digit. The notational system of the secret digits can be every desired number, as well as the size of the pixel groups. In order to preserve the quality of the stego image, it is desired to limit the changes in the pixel groups as much as possible. Therefore, optimum strategies must be found to match the function of the pixel group to the secret digit with the least possible imposed distortion in terms of mean square error. Having been recently found for pixel pairs, such strategies are found for the larger pixel groups by the proposed method in this paper. Among all the strategies providing the similar minimum MSE value, the one is chosen that helps to preserve the histogram of the original image. Optimum strategies found for all notational systems and pixel group sizes makes the algorithm flexible for various application with different payloads, while it improves the similar techniques in terms of both MSE reduction and histogram preservation, as is confirmed by the experimental results.

component, adaptive pixel group matching (APGM), exploiting modification direction (EMD), least significant beats (LSB) steganography.