In this study, a new steganalytic method, which exploits the difference statistics of neighboring pixels, is proposed to detect the presence of spatial LSB matching steganography. In the proposed method, the differences between the neighboring pixels (DNPs), the differences between the local extrema (DLENs) and their neighbors in grayscale histogram are used as distinguishing features and the SVM is adopted to construct classifier. Experimental results show that the proposed method is efficient to detect the LSB matching steganography for the compressed and uncompressed images and outperforms other recently proposed algorithms.
展开▼
