Cryptanalysis of a wavelet based watermarking scheme

In this paper we cryptanalyze the wavelet based watermarking scheme by Wang et al, 2002, By cryptanalysis we mean the removal of the watermark using a single watermarked copy which is equivalent to cipher text only jamming attack. The watermark embedding process replaces a particular middle frequency band of the host image (in wavelet domain) with the watermark. In the scheme, the key consists of three secrets: (i) the watermark, (ii) the random filter bank used for wavelet transform and (iii) the wavelet band where the watermark is inserted. First we observe that the secret random filter bank does not provide any security since it can be replaced by any filter bank from a large class. Further, it is possible to discover the secret wavelet band used to watermark the host image. Though in cryptography the random nature of secret key provides the security, in contrary here we show that the random nature of the watermark actually helps in identifying the secret wavelet band and consequently one can remove the watermarking signal from that band to mount the successful cryptanalytic attack.