Embedding ratio estimation based on weighted stego image for embedding in 2LSB

For two different steganography paradigms of 2LSB steganography, two corresponding steganalysis methods are proposed based on weighted stego image. This paper gives and improves two theorems that for two cases of 2LSB steganography, when the weight parameters are equal to the embedding ratios, the weighted stego image is closest to the cover image in the least square sense. Afterward, based on the two theorems and a predictor of cover image, two steganalysis methods are proposed to estimate the embedding ratios. Experimental results show that the new steganalysis methods outperform the structural steganalysis when the embedding ratio in any bit plane is middling or large. And this can fetch up the defect of structural steganalysis.