Abstract:
This paper presents a new method for high-capacity robust watermarking of hyperspectral images. A new algorithm for two-dimensional “noise-like” watermarking patterns generation is proposed as part of robust watermark embedding and detection procedures. The experimental research provided in this work shows that the average complexity of the brute-force key retrieval attack can be increased to $10^{14}$ watermark extraction attempts (compared to $10^4$–$10^5$ for known robust watermarking schemes). Experimental results also show that the watermark preserves its robustness
against lossy compression of host video, at the same time preserving a higher video quality (PSNR
up to 51dB) compared to known wavelet-based and DCT-based watermarking algorithms.