Improved steganalysis of LSB matching steganography based on counting alteration rate of the number of neighbourhood gray levels

Abstract

This paper proposes an improved method for attacking the LSB (Least Significant Bit) matching steganography. In LSB matching steganography the least two or more significant bit-planes of the cover image would be changed during the embedding and thus the pairs of values do not exist in stego image. In the proposed method, an image is obtained by combining the least three significant bit-planes and is divided into 3x3 overlapped subimages. The subimages are grouped into eight types, i. e. T1, T2, T3, T4, T5, T6, T7 and T8 according to the count of gray levels. A message is embedded by LSB matching. The alteration rate of the number of elements belonging to T1 is computed. It is found that normally the alteration rate is higher in cover image than in the corresponding stego image. This is used as the discrimination rule in the method. The literature consists of the method which focuses on the combination of least two significant bit-planes. Experimental results demonstrate that the proposed method gives better results than those obtained by combining the least two significant bit-planes. It is an efficient method to detect the LSB matching stegonagraphy in gray scale images.