摘要
针对当前遥感图像融合算法在全色图像与多光谱图像融合过程中存在光谱失真和模糊等问题,提出一种基于卡通-纹理分解的融合方法。通过特征分解将遥感图像分解为具有光谱信息的卡通空间和包含空间信息的纹理空间;采用全色纹理分量替换多光谱纹理分量,对遥感图像进行全色锐化;利用全色卡通元素的梯度,在保持光谱质量的同时,强化图像的边缘信息和一些重要细节。实验结果表明,该方法可以在保留光谱特征的同时获得高分辨率的多光谱图像,而且该方法在空间和光谱质量方面的性能表现明显优于其他方法。
Aiming at the problems of spectral distortion and ambiguity in the fusion process of the panchromatic image and multi-spectral image,this paper proposes a new fusion method based on cartoon-texture decomposition.The remote sensing image was decomposed into cartoon space with spectral information and texture space with spatial information by feature decomposition;the panchromatic texture component was replaced by the multispectral texture component to sharpen the remote sensing image;the gradient of panchromatic cartoon elements was used to enhance the edge information and some important details while maintaining the spectral quality.The experimental results show that the proposed method can obtain high-resolution multispectral images while retaining spectral features,and its performance in spatial and spectral quality is significantly better than other methods.
作者
潘多
周雪梅
Pan Duo;Zhou Xuemei(Department of Information Engineering,Sichuan Technology Business College,Dujiangyan 611830,Sichuan,China)
出处
《计算机应用与软件》
北大核心
2020年第9期165-169,175,共6页
Computer Applications and Software
基金
教育部科技发展中心产学研创新基金课题(2018A03007)。
关键词
卡通-纹理分解
全色锐化
图像分解
遥感图像融合
Cartoon-texture decomposition
PAN-sharpening
Image decomposition
Remote sensing image fusion
