In this paper, security evaluation of an important watermarking class based on quantization is given. Theoretical developments and practical simulations are used to measure the security level of watermarking techniques. We give the contribution of each observation available to the attacker on the total gathered information about the watermarking secrecy. By leading on watermarking technique weaknesses, we find that using the Quantization Index Modulation (QIM) with continuous secret key is equivalent to using a secret key with two states. The latter is easier to estimate than a continuous key. Then, we propose a secure version of a classical trellis coded quantization watermarking. The security is guaranteed thanks to the trellis path generated from the discrete key and the message. We show that the spread transform can represent a second or alternative security level for watermarking systems. It allows to increase the watermarking security level and to keep the embedded message hard to read for unauthorise user.Index Terms-Security.
INTRODUCTIONMany of Watermarking techniques claim to be secure just because a secret key is used at the encoder. In reality, they present many weaknesses and the watermarking protection can be broken. The security of a watermarking concerns its capability resist to intentional forgeries. In [1], the authors have proposed the foundations of the watermarking security measure. By leading on Shannon theory on cryptographic security presented in the famous paper "Communication Theory of Secrecy Systems" [2], they evaluate the security of watermarking system by measuring the information leakages about the secret key, used to prevent the unauthorized users reading the embedded message, from the observations available to the attacker. Roughly, the information leakage is calculated thanks to the mutual information between the observation (watermarked copie) and the discrete secret key. When the key is continuous, the information leakage is measured thanks to the fisher information about the secret key from the observation. In both case, we measure the amount of observations (watermarked copies) needed by the attackers to gather enough information and recover the secret key used at the encoder. This paper focuses on the security of an important class of watermarking schemes based on quantization. A security of the distortion compensated (DC)-Quantization Index Modulation (QIM) or DC-QIM[3] -equivalent to the Scalar Costa Scheme (SCS)[4]-has been studied in [5] "practically", when a continuous key is used. However, studying the security against specific (practical) attack does not give the limit of the watermarking system security. Luis Pérez-Freire