基于改进视觉模型的自适应图像水印算法

Adaptive Image Watermarking Algorithm Based on Improved Perceptual Models

摘要: 基于量化的水印嵌入算法可以实现盲检测，QIM(Quantized Index Modulation)是最常见的量化嵌入方法。量化步长是影响量化水印算法性能的最重要因素之一。本论文基于视觉模型的特点，针对多种具体的攻击，提出了对视觉模型进一步改进以及改进视觉模型下的四种不同水印嵌入算法，并将其与QIM相结合。实验结果表明本论文提出的算法对噪声干扰和常见的图像处理均具有较好的鲁棒性。论文最后给出总结和展望。

Abstract: A blind detection can be achieved based on the quantization of the watermark embedding algorithm. QIM (Quantized Index Modulation) is one of the most common quantization embedding methods. The quantization step is one of the most important factors which affect the performance of quantization watermarkings. In this paper, according to the characteristic of perceptual model and a variety of attacks, further modified perceptual model and different imple- mentations of perceptual model are proposed. They are incorporated with the spread transform quantization index modulation (ST-QIM) framework. The experimental results show that the four algorithms we proposed in this paper are robust to noise attacks and common digital image processing operations. Finally, in conclusion section, summary and outlook are given.