VLSI Implantation of Image Watermarking using TIWT with Hybrid Accumulation

The VLSI implementation of digital watermarking system must meet the low area (look up tables, slices, registers, memory elements), delays (path, logical) delays. However, the conventional discrete wavelet transforms (DWT) based VLSI watermarking systems suffers with higher consumption of these attributes due to down sampler. So, this work is focused on implementation of translation invariant wavelet transform (TIWT) based watermarking system using VLSI fundamentals. The grayscale image is decomposed into several frequency bands using the TIWT algorithm. This results in a set of coefficients representing the image in each frequency band. A subset of low coefficients is selected from one or more of the frequency bands for embedding the watermark. The selected coefficients are modified to embed the watermark. The modification is performed using a pseudorandom sequence, which is generated using a secret key. The sequence determines the locations and values of the watermark bits that will be embedded in the selected coefficients. Here, embedding operation is implemented using VLSI based adders and multipliers, such as hybrid accumulation. Finally, the modified coefficients are used to reconstruct the watermarked image using the inverse TIWT algorithm. The simulation results show that, the proposed TIWT system resulted in reduced hardware resource utilization, and improved watermarking robustness as compared to existing methods.