在小波包域提出了一种多传感器图像融合和双水印算法。首先, 利用HIS变换和小波包变换将多光谱和全色图像分解为多个高低频子带, 根据小波包域系数特点, 低频部分采用基于区域平均能量加权算法的规则进行融合, 高频部分采用绝对值取大的规则进行融合。然后, 在高低频图像融合系数分别嵌入一个水印, 低频水印利用了离散余弦变换的聚能去相关能力, 高频水印利用了图像纹理子块特征。最后, 对嵌入的双水印融合图像进行攻击和分析。实验结果显示, 融合图像在保留多光谱图像光谱信息的基础上有效提高了空间分辨率; 加水印的融合图像具有良好的不可视性和鲁棒性。

A method of multisensor image fusion and bi-watermarking based on Wavelet Packet Transform(WPT)is proposed in this paper. First, the multispectral and panchromatic images are decomposed into more different frequency subbands by HIS and WPT transform. Due to the properties of the WPT coefficients, a regional energy weighting image fusion algorithm is presented for the low frequency subband coefficients, and for the high frequency subband coefficients, the algorithm uses the absolute value maximum selectivity scheme. Then, the watermarks are embedded in the low and high frequency of the fusion image wavelet packet decomposition. The low frequency watermark uses the DCT energy compression ability and the ability of ridding correlation, and the high frequency watermark uses the streak blocks of the fusion image. Finally, the watermarked fusion image is attacked. The experimental results show that, the watermarked fusion image keeps good spectral information and spatial resolution, and the watermarks embedded in the fusion image are robust and invisible.