Detect LSB Steganography with Bit Plane Randomness Tests

We present a novel steganalytic technique for images that have been embedded hidden messages with LSB type steganographic algorithms. Two binary sequences is obtained by scanning the seventh and eighth bit planes of the image with Hilbert scan, then the randomness of this two sequences are tested by several randomness tests respectively. The key idea is that, the randomness of the bit planes will differ between a stego-image and a cover-image. The results of the randomness tests are used as image attributes to construct a classifier to distinguish between stego and cover images. The experiments show the detection accuracy of our method is higher than 95% to stego images with a embed rate higher than 0.05 bits per pixel. The experiments also show our method is effective to some frequency domain steganographic algorithms such as F5