摘要
为了使融合后的图像在保持原IKONOS卫星图像多光谱特性的同时,最大可能地提高图像空间分辨率,提出了一种基于四树复小波包变换的SAR图像与多光谱IKONOS卫星图像相融合的新方法.该方法利用复小波包变换的多方向性和对高频细节信号良好的时频局部化分析能力,分别对IKONOS图像经HIS空间变化的I分量子图和SAR图像进行复小波包分解,并对分解后的低频复系数采用取平均或取大值的方法、对高频方向复系数采用邻域一致性测度的局部自适应方法进行复系数融合.用融合后复系数经复小波包反变化得到的图像代替原IKONOS图像经HIS变换的I分量,再经HIS空间反变换得到最终的融合图像.实验结果表明,该融合算法在光谱保留和空间质量提高方面,比传统的基于小波变换的融合算法具有更高的性能.
In order to enable the fused image to maintain multi-spectral characteristics of the original IKONOS satellite images and improve the image space resolution to its utmost limits, a new image fusion method based on quad-tree complex wavelet packet transform (QCWPT) is presented for multispectral IKONOS satellite image and SAR (Synthetic Aperture Radar) image, and the proposed image fusion method employs the multi-directional characteristics and excellent local time- frequency analysis ability of QCWPT for high frequency detail signals. The I component sub-image of the multispectral IKONOS satellite image via HIS space transform and the SAR image are decomposed separately by using the QCWPT. The low frequency complex wavelet packet coefficients are fused by using an average or large-value fusion method, and the high frequency complex wavelet packet coefficients are fused via a new fusion method, namely, a locally adaptive fusion method for neighborhood homogeneous measurement. The original i component sub-image of the multispectral IKONOS satellite images is replaced by the reconstruction image via reverse QCWPT. So, the last fusion image can be obtained via HIS space reverse transform. Experimental results show that the presented fusion algorithm is of higher performance than other traditional fusion methods in spectrum reservation and space quality improvement.
出处
《信息与控制》
CSCD
北大核心
2008年第4期396-402,共7页
Information and Control
基金
重庆市科委自然科学基金资助项目(CSTC:2006BB2393)
浙江省教育厅科研项目(20061661)
关键词
图像融合
四树复小波包变换
SAR图像
多光谱IKONOS卫星图像
邻域一致性测度
image fusion
quad-tree complex wavelet packet transform (QCWPT)
SAR (Synthetic Aperture Radar) image
multispectral IKONOS satellite image
neighborhood homogeneous measurement