基于邻域匹配度和能量的curvelet变换融合

提出一种基于邻域匹配度和能量加权平均的多源遥感图像曲波变换融合方法。对低分辨率的多光谱图像和高分辨率的全色图像作曲波变换,融合图像的曲波系数中的低频分量取自多光谱图像的低频分量。求两幅图像高频系数邻域内的匹配度,与阈值相比较,根据比较结果分别采用不同方法确定高频系数。对生成系数进行逆曲波变换后得到融合图像。实验表明融合后的图像清晰度和光谱性得到有效提高。

Curvelet transform for image fusion based on matching degree and energy of neighboring region

An efficient method of image fusion based on Energy and matching degree of curvelet coefficients neighboring region is proposed in this paper.Firstly,the multi-spectral and high resolution images are decomposed by curvelet transform.Then the curvelet coefficients of the fused image can be obtained by using different rules.The approximate coefficients come from the multispectral image,and the detail coefficient is based on the matching degree and Energy of neighboring region.Finally,the fused image is obtained by inverse curvelet transform.The experiment results show that the average gradient and preserve spectrum index commendably improved a lot.