摘要
当前遥感图像融合方法忽略了不同图像像素间的关联性,导致融合图像含有伪吉布斯效应及光谱失真,为此提出了二代Curvelet变换耦合细节度量模型的遥感图像融合算法。引入HSV变换来分解多光谱(MS)图像,提取相应的亮度成分。借助二代Curvelet变换来分解亮度成分与全色(PAN)图像,获取其不同的子带。构造低频子带融合准则,获取对应的权重系数,实现低频子带的融合。对高频子带实施卷积操作,提取其边缘信息,将其与空间信息相结合,构造细节度量模型,对高频子带的边缘及空间细节特征进行度量,实现高频子带的融合。利用二代Curvelet逆变换对融合子带进行计算,获取新的亮度成分,并将其与初始的色度、饱和度信息进行HSV逆变换,实现遥感图像融合。实验结果显示,较当前的遥感图像融合方法而言,所提算法具有更高的融合质量,其输出图像不仅清晰度更高,而且也有更好的光谱特性。
In order to solve the problem that false Gibbs effect and spectral distortion in the fused image induced by neglecting the correlation among different image pixels in many remote sensing image fusion methods,a remote sensing image fusion method based on second-generation Curvelet transform coupled detail measurement model is proposed.HSV transform is introduced to decompose MS images and extract brightness factors.The second generation Curvelet transform is used to decompose the brightness factor and PAN image for obtaining different sub-image information.The fusion factor of low-frequency subgraphs is constructed,and the weight coefficients in the process of fusion of low-frequency subgraphs are obtained to realize the fusion of low-frequency subgraphs.The edge information of high-frequency subgraph is extracted by convolution operation,and the edge information of high-frequency subgraph is combined with spatial information to construct a detail measurement model,which measures the edge and spatial details of high-frequency subgraph and realizes the fusion of high-frequency subgraph.second-generation Curvelet inverse transform is used to calculate the fused sub-image,obtain new brightness factor,and perform HSV inverse transform with the chroma factor and saturation factor to realize remote sensing image fusion.The experimental results show that the fused image of the proposed method not only has higher clarity,but also has better spectral characteristics and visual effect than the current method.
作者
罗娟
王立平
谭云兰
Luo Juan;Wang Liping;Tan Yunlan(Yichun Early Childhood Teachers College,Gaoan 330800,China;College of Mechanical and Electronic Engineering,Pingxiang College,Pingxiang 337055,China;College of Electronic and Information Engineering,Jinggangshan University,Ji’an 343009,China)
出处
《电子测量与仪器学报》
CSCD
北大核心
2019年第7期129-136,共8页
Journal of Electronic Measurement and Instrumentation
基金
江西省重点研发计划(20171BBE50049)
江西省知识产权软科学研究计划(ZR201610)
江西省高校人文社会科学重点基金(JD17127)资助项目