Multispectral and panchromatic image fusion using multi-feature and wavelet transform

作者：Luo Xiaoqing^*; Wu Xiaojun

来源：International Review on Computers and Software, 2012, 7(5): 2034-2040.

• Component analysis
• First principal components
• Frequency coefficient
• Fusion image
• Fusion methods
• Fusion rule
• High frequency
• High resolution
• High spectral resolution
• Multi-feature
• Multi-spectral
• Multispectral images
• Numerical evaluations
• Panchromatic images
• Remote sensing images
• Spectral aspects
• Statistical models
• Subbands

It is an important issue in remote sensing image fusion that effectively combines the high spatial and high spectral resolutions in order to obtain the complete and accurate description of the observed scene. In previous research, various algorithms have been proposed. However, the available methods could hardly produce the satisfactory results especially in the spectral aspect. In this paper, a new fusion method based on multi-feature and wavelet transform is proposed to fuse panchromatic image (PAN) with multispectral image (MS). This approach uses entropy component analysis (ECA) and adaptive principal component analysis (APCA) to eliminate band redundancy. The first principal component of MS (FPCM) and PAN is decomposed using wavelet. PAN is divided into windows and extract multi-feature of window and then get region. For low frequency subband, the low frequency coefficient of FPCM is selected as that of fusion image. A statistical model method based on region is adopted as high frequency fusion rule. High resolution multispectral image is then obtained by an inverse wavelet and ECA transform. Experimental results including visual and numerical evaluation also proves the superiority of the proposed method to its counterparts.

• 出版日期2012
• 单位江南大学