Secure hybrid robust watermarking resistant against tampering and copy-attack

Abstract

The present invention relates to the methods for hybrid watermarking method joining a robust and a fragile watermark, and thus combining copyright protection, authentication and tamperproofing. As a result this approach is at the same time resistant against the copy attack. In addition, the fragile information is inserted in a way which preserves the resistance and reliability of the robust part.

We claim:  
     
         1 . A method for generating watermarked data z based on some original data x, wherein said robust watermark w contains multi-bit informative w inf  and reference w ref  watermarks encoded and embedded in such a way as to resist against attacks, and an authentication watermark w frag  contextually encoded, encrypted and embedded in orthogonal or almost orthogonal positions with respect to the robust watermark w, comprising the steps of: 
 (a) encoding said multi-bit message b, possibly using any error correction code (ECC),    (b) generating said w as a function of said key k and said message b encoded and/or encrypted as codeword c, where said w consists in a said informative watermark w inf  and said reference watermark w ref ,    (c) generating said authentication watermark w frag  as a function of said key k and said contextual information,    (d) embedding said robust watermark w into said original data x to get said robustly marked data y,    (e) embedding said authentication watermark w frag  into said robustly marked data y in orthogonal manner to said informative watermark w inf , resulting into the final marked data z    whereby said reference watermark w ref  assists amongst others in the estimation and recovering from local and global geometrical image alterations, channel state estimation, verification of reliability, fast detection of the said robust watermark w presence in said z and synchronized decoding of said error correction codes (ECC), and said authentication watermark w frag  assists in tampering detection, authentication and prevention of estimation of robust watermark w or informative watermark w inf  with following copying to another target media known as the copy attack, and identification of the reliability of said informative watermark w inf , and furthermore whereby said z is visually indistinguishable from said x.    
     
     
         2 . The method of  claim 1  wherein said function uses perceptual masking M while adding said watermark w and said authentication watermark w frag  to said x in the spatial domain or some transform domain.  
     
     
         3 . The method of  claim 1  wherein said authentication watermark w frag  contains global and/or local encrypted contextual information about data x such as data size, unique data ID, name, index or random unique stamp that is the same for one data but is different to another data to resist against collage attack and copy attack, and produces local data dependent signatures or local message digest code (MDC).  
     
     
         4 . The method of  claim 1  and  3  wherein said unique stamp additionally carries information about date, time and other specific information identifying the particularities of embedding process that even enables the identification of copied areas and their localization.  
     
     
         5 . The method of  claim 1  wherein said encoded and encrypted authentication watermark w frag  uses as input global and local information about the original data x and/or the robustly watermarked data y, including local blocks of the data, local blocks indexes, data global size, etc., and wherein all key dependent positions and bit planes where said authentication watermark w frag  is to be embedded are excluded from the information used as input for the generation of this w frag .  
     
     
         6 . The method of  claim 1  wherein said encoded and encrypted authentication watermark w frag  comprises a key-dependent regular (such as square blocks), or any irregular spatial allocation structure.  
     
     
         7 . The method of claims  1 ,  5  and  6  wherein said blocks are hashed using information about neighboring blocks to defeat simple substitution attacks and cut-and-paste attacks.  
     
     
         8 . The method of claims  1  wherein said authentication watermark w frag  is embedded into contiguous and non-overlapping blocks with predefined indexes which may be generally key-dependent.  
     
     
         9 . The method of  claim 1  wherein said informative w inf  and authentication watermarks w frag  are used jointly to detect the collage and the copy attacks that can not be achieved only based on their independent usage.  
     
     
         10 . The method of  claim 1  and  9  wherein said informative w inf  and authentication w frag  watermarks are used jointly on the local blockwise level to detect the sequence of the applied attacks and to distinguish the multiple watermarks embedded with the same technology and the same key but possibly with different messages, and to identify the original informative message b initially embedded into said media x.  
     
     
         11 . The method of  claim 1  wherein said informative watermark w inf  and/or authentication watermark w frag  is used to detect the tampered regions or the boundaries of objects, which have been copied from other medias even with the preservation of their block indexes and positions.  
     
     
         12 . The method of  claim 1  wherein said original data x is video, audio or image data.  
     
     
         13 . The method of  claim 1  applied to video data, wherein a plurality of watermarked video frames is generated.  
     
     
         14 . The method of  claim 1  wherein said function operates in the spatial domain, Discrete Cosine Transform (DCT) domain, Discrete Fourier Transform (DFT) domain, wavelet domain, or any other transform domain, or some combination thereof.