In 2003, Chu proposed an oblivious watermarking algorithm by modifying the CKLS scheme proposed by Cox, Kilian, Leighton, and Shamoon in 1997, known as the CKLS scheme. In this correspondence, we report that the modification presented by Chu is susceptible to a suitably modified attack devised by Das and Maitra in 2004. In fact, the experimental results show that Chu's scheme is even weaker than the CKLS scheme in terms of our attack.

Most of the existing digital watermarking techniques are based on correlation between "some information stored in the watermarked copy" and "related information retrieved from attacked watermarked copy". We show how to remove this correlation to mount a ciphertext-only cryptanalytic attack on these watermarking schemes. The attack needs only a single watermarked copy of the image. Our work clearly raises question on the basic watermarking paradigm based on correlation measures.

In invisible digital watermarking, given an image I, a signal s/sub i/ is added, which produces a watermarked copy I/sup (i)/ = I + s/sup (i)/, such that I, I/sup (i)/ are visually indistinguishable. The addition means the element wise addition in the image matrix. This image I/sup (i)/ is given to the i-th buyer. In the detection stage, the available image (may be attacked using image processing. or cryptanalytic techniques) I/sup (#)/ is compared to the original image I and a signal s/sup (#)/ = I/sup (#)/ - I is recovered. The buyer i is suspected if s/sup (i)/ possesses significant correlation with s/sup (#)/. For successful cryptanalysis, one has to mount an attack to construct I/sup (#)/ from I/sup (i)/ (both visually indistinguishable) such that there is no significant correlation between s/sup (#)/ = I/sup (#)/ - I and s/sup (i)/. Thus, the buyer i will not be identified. To the attacker, only I/sup (i)/ is available, but I, s/sup (i)/ are not known. Thus there is no facility for the attacker to test the correlation between s/sup (i)/ and s/sup (#)/. However, the attacker needs to be convinced indirectly that the correlation between s/sup (i)/ and s/sup (#)/ has been removed. We use the CKLS scheme to demonstrate the success of the attack. The paradigm of attack needs only a single copy in contrast to the collusion attack, which needs more than one copy.

This paper proposes a novel approach to automatic recognition of handprinted Bangla (an Indian script) numerals. A modified Topology Adaptive Self-Organizing Neural Network is proposed to extract a vector skeleton from a binary numeral image. Simple heuristics are considered to prune artifacts, if any, in such a skeletal shape. Certain topological and structural features like loops, junctions, positions of terminal nodes, etc. are used along with a hierarchical tree classifier to classify handwritten numerals into smaller subgroups. Multilayer perceptron (MLP) networks are then employed to uniquely classify the numerals belonging to each subgroup. The system is trained using a sample data set of 1800 numerals and we have obtained 93.26% correct recognition rate and 1.71% rejection on a separate test set of another 7760 samples. In addition, a validation set consisting of 1440 samples has been used to determine the termination of the training algorithm of the MLP networks. The proposed scheme is sufficiently robust with respect to considerable object noise.