基于二维不可分形态小波变换的多光谱图像全色锐化方法

A multispectral pansharpening method based on 2D nonseparable morphological wavelet transform

针对全色锐化中如何有效提取高分辨率细节这一关键问题,提出了一种新的基于二维不可分形态小波变换的全色锐化方法.结合了线性小波与形态小波的各自优势,在保持小波消失矩的同时能够有效保留图像的边缘细节.实验部分采用SPOT 6与Pléiades两组遥感卫星图像数据.实验结果证实本文所提方法能够降低光谱误差,并有效提升了多光谱图像的空间分辨率,在ERGAS、Q4等评价指标方面优于当前主流全色锐化算法,同时取得了良好的视觉增强效果.

A new multispectral pansharpening method based on 2D nonseparable morphological wavelet transform is proposed in this paper, aiming at solving the key issue that how to extract the high resolution details effectively. The proposed method combines the advantages of linear and morphological wavelets, which can retain the edge details of images while maintaining the wavelet vanishing moment. Two spaceborne image data are taken including SPOT 6 and Pleiades. Experimental results show that the proposed method can effectively reduce the spectral distortion while improving the spatial resolution of multispectral images. The performance in terms of eval uation indexes such as ERGAS, Q4 outperforms the state of the art pansharpening algorithms, and good visual enhancement effect is achieved.