论文部分内容阅读
针对遥感影像的安全保护,提出了一种基于SURF特征区域的鲁棒水印算法。首先在影像中检测SURF特征点,选取鲁棒性强的点构造互不重叠的特征区域并进行归一化处理,进而对归一化的区域实施整数小波变换,根据生成的模板选取相应位置的系数,最后通过量化的方式将水印信息重复嵌入到每个特征区域内。水印的检测可直接在攻击后的影像中进行,无需校正恢复,实现了盲提取。采用边缘像素替换的策略来保证影像的数据精度,不影响遥感影像的使用。实验表明,该算法具有良好的不可见性,且对常规攻击和几何攻击均具有较强的鲁棒性,可有效保护遥感影像的安全。
Aiming at the security protection of remote sensing images, a robust watermarking algorithm based on SURF feature region is proposed. Firstly, the SURF feature points are detected in the image, the strong points are selected and the non-overlapping feature regions are constructed and normalized, and then the integer wavelet transform is performed on the normalized regions. Based on the generated template, the corresponding position Coefficient, and finally, the watermark information is repeatedly embedded into each feature area by the way of quantization. Watermark detection can be carried out directly in the image after the attack, without correction and recovery, to achieve a blind extraction. Edge pixel replacement strategy to ensure the accuracy of the data of the image, does not affect the use of remote sensing images. Experiments show that the algorithm has good invisibility and robustness against both conventional attacks and geometric attacks, which can effectively protect the safety of remote sensing images.