Detection of random LSB image steganography

The spatial LSB steganography results in the alteration of the smooth characteristics between adjoining pixels of the raw image. The relation between the length of embedded message and the gradient energy is theoretically analyzed, and then a steganalysis and detection method, named gradient energy-flipping rate detection (GEFR) is proposed. Through the analysis of the variation of the gradient energy, which results from the LSB steganognaphy in color and grayscale images, the secret message embedded in the target image is detected, and the length of the embedded message is estimated. The method is proved effective and accurate by simulation (detection rate reaches 0.01 bit per pixel). The applicability of all kinds of images to this algorithm is analyzed, and the error emendation method for special kinds of images is presented.