摘要： 针对Laplace算子在进行边缘检测时不具有方向性的缺陷，在人眼视觉模型下，提出一种基于边缘检测的半脆弱水印算法。在嵌入水印信息时先考虑边缘点，再考虑亮度较暗和较亮区域。在人眼亮度和纹理视觉特征的基础上，给出图像区域分割方法及变换域系数量化公式，通过动态量化提高载体图像的透明性和水印的抗攻击能力。实验结果表明，在噪声嵌入量增大的情况下，该算法的透明性及顽健性均优于传统半脆弱水印算法。

关键词: 数字水印, 半脆弱水印, 像素值调整, 边缘检测, Laplace算子

Abstract: Laplace operator has not direction performance when conducting edge detection. Aiming at this problem, a semi-fragile watermark algorithm based on edge dectation is given under the eye vision model. It considers edge points and dark or light area when embedding watermark information. On basis of visual features of brightness and texture in the human eye, a method of image region segmentation and transform domain coefficient quantization formula are introduced to improve the image watermark transparency through dynamic vector quantization and anti-attack capability. Experimental results show that, with the amount of embedded case increasing, the algorithm of transparency and robustness are higher than the current semi-fragile watermarking algorithm.

Key words: digital watermark, semi-fragile watermark, pixel value adjustment, edge detection, Laplace operator

中图分类号: 

• TP311.52