Article Preview
TopInformation hiding algorithms by changing geometric features can be divided into algorithms in spatial domain and algorithms in frequency domain. In the spatial domain, a scale digital watermarking algorithm for 2D engineering graphics is proposed in Wang and Xu (2003). It uses vertices of all entities as watermark carriers and can resist attacks such as translation, rotation and scaling. However, it is vulnerable to entity addition and deletion. A spread spectrum digital watermarking algorithm for 2D engineering graphics is proposed in Wang, Liu, and Xu (2005). Firstly, the narrow band signal for embedding is spread into wide band pseudo-random sequence, then the watermark is embedded by modulating and modifying the length of the lines. It is robust against the translation, rotation, scaling, entity addition and deletion, but the capacity is small and spread spectrum synchronization is complex.
In the frequency domain, a non-blind detection watermarking algorithm for 2D engineering graphics is proposed in Huang and Gu (2006a), the vertices of all entities are acted as watermarking carriers. Firstly, the coordinates of 2D engineering graphics are used to construct complex signals, and then Fourier transform is applied to them. After that, the watermark is embedded into the magnitude of the Fourier coefficients. Due to the advantages of Fourier transform, it can resist attacks such as translation and rotation. Since the embedded watermark information is redundant, it is also robust against attacks such as entity addition and deletion, but cannot resist against scaling. A watermarking algorithm based on chaotic and wavelet transform for 2D engineering graphics is proposed in Huang and Gu (2006b). The coordinates of vertices are used to construct complex signals, and wavelet transform is applied to the complex signals. After that, the watermark encrypted by chaos sequence is embedded by modulating the amplitudes of detail wavelet coefficients. Because of the characteristics of wavelet transform, it is robust against translation, rotation and scaling. It can also resist attacks such as entity addition and deletion, but the embedding capacity is relatively low.
After an analysis of the information hiding algorithms for 2D engineering graphics based on geometric features, we find that the essence of them is that the watermark embedded is based on the modification of the vertex coordinates. Here, a steganalysis method is proposed based on this principle.