一种基于区域分割的多尺度遥感图像融合方法

光谱保持和高分辨率保留是图像融合的重要问题,提出了一种区域分割和小波变换相结合的多尺度遥感图像融合方法。首先对经过配准的待融合图像进行小波变换,然后对变换后的低频系数进行基于区域标准差的分割,将低频系数分为目标信息和背景信息,接着对目标信息采取基于绝对值的融合,对背景信息采用基于灰度误差的融合。对小波变换后的高频系数采用基于清晰度的融合规则,最后进行小波逆变换得到融合图像。将该方法和几种常用融合方法进行对比分析,结果表明:该方法在有效地保持多光谱影像光谱信息的同时,可以有效地提高融合影像的空间细节信息,有利于后续进行信息提取和图像分类。

Multi-scale Remote Sensing Image Fusion Method based on Region Segmentation

Retaining spectral and high resolution are the important issue of image fusion.An image fusion method based on Region Segmentation and Wavelet transform for multi-scale remote sensing image fusion is proposed.Firstly,images must be corrected,then the multiscale decomposition of source images is carried out with the wavelet transform.Then region segmentation based on area standard deviation is done for the low-frequency coefficients,which is decomposed two parts:target information and background information.The target information is fused with larger absolute value operator,and the background information is fused with gray error value operator.The high-frequency coefficients are fused with image definition.Finally the fused coefficients are reconstructed to obtain the fusion image.Compared with several traditional methods,the results show that the fused image by the presented algorithm can not only hold spectrum information of the multispectral image,but also improve the high resolution of the fusion image,then preparing for information extraction and image classification.