摘要
针对高分一号(GF-1)遥感影像高空间分辨率的特点,综合压缩感知理论改进了传统的IHS影像融合算法,利用稀疏基和测量矩阵对多光谱影像IHS变换后的I'分量和全色影像进行处理,采用加权平均和OMP(orthogonal matching pursuit)重构得到新的I分量,再通过IHS反变换得到结果影像,并结合5个定量指标进行分析评价。实验结果表明,与传统方法相比,结合压缩感知的IHS融合算法所得相关系数更高、扭曲程度更小,融合结果不仅具有更高的空间信息丰富度,并且保持了多光谱影像的色彩信息,可为GF-1影像的目视解译和影像分类提供参考。
According to characteristics of GF-1 remote sensing images with high spatial resolution,the authors used compressed sensing theory to improve the traditional IHS image fusion algorithm. The component I from IHS transform and panchromatic images used sparse matrix and measure matrix,the weighted average and OMP yielded new component I＇. Finally,through an inverse IHS transform the result image was obtained. Combined with five quantitative indexes,analysis and evaluation were conducted. Experimental results show that,compared with the traditional methods,IHS fusion algorithm combined with compression perception can obtain a higher and less distorted correlation coefficient,and the fusion results not only have higher spatial information richness,but also maintain the color information of multi-spectral images. It may provide a reference to GF-1 image visual solutions for translation and image classification.
出处
《国土资源遥感》
CSCD
北大核心
2017年第4期26-32,共7页
Remote Sensing for Land & Resources
基金
国家自然科学基金项目"高分辨率SAR图像典型地物目标样本特征提取和识别研究"(编号:61372189)
中国科学院大学生创新实践训练计划项目"基于压缩感知的遥感影像融合研究"(编号:Y5Y01206QM)共同资助
关键词
影像融合
IHS变换
稀疏基
测量矩阵
OMP重构
remote sensing image fusion
IHS transform
sparse basis
measurement matrix
OMP reconstruction
