Secure Encryption and Hiding of Intelligence Data.

Abstract : A method for hiding messages in the noise component of digital images in such a way that specific statistical characteristics of the carrier image are not disturbed is presented. The bits are injected in the secret message, taking into account the properties of the carrier image. Adaptive message hiding should be based on local properties of the carrier image. Secure steganographic techniques using digital images require message bits to be spread pseudo-randomly over the carrier image independently of the message lengths and not concentrated in the first rows or columns of the image. For this purpose, a simple but effective message-spreading algorithm based on parameterized chaotic maps was developed. The algorithm uses discretized two-dimensional chaotic maps as universal dispersers for random message scattering within the carrier image. This reports describes several new designs for secure watermarking schemes for digital images. The techniques offer a very high degree of robustness and security with respect to image modifications including noise and lossy compression. A technique is presented for overlaying the carrier image with a key-dependent image whose power is concentrated in the low frequencies. Message extraction is based on a discrete cosine transform. A second technique inserts bits into projections of image blocks onto random, smooth, orthogonal patterns individually generated for each image and user ID. This method avoids using publicly known basis functions, thereby increasing security against malicious attacks. Robustness test results against image distortion and manipulation techniques and security of attacks to removing hidden messages are presented. Patent application serial number PASN 08/986695, 8 Dec 97, Title "Digital Watermark" has been filed for this work.