Acta Optica Sinica, Volume. 33, Issue 4, 428003(2013)
Remote Sensing Image Fusion Based on Sparse Representation
In order to improve multi-spectral (MS) image fusion quality, a new pan-sharpening method based on sparse representation is proposed. A linear regression model between the MS image and its intensity component is established. The sparse coefficients of both panchromatic image and MS image are obtained by two dictionaries which are trained to have the same sparse representations for each high-resolution and low-resolution image patch pair. The coefficient of intensity can also be obtained via the linear regression model and the coefficients of MS bands. Then, the sparse coefficients are fused in the general component substitution (GCOS) fusion framework. The fused sparse coefficients are used to reconstruct a high-resolution MS image. As the inherent characteristics and structure of signals are by via sparse representation more efficiently, the proposed method can preserve spectral and spatial details of the source images well. Experimental results on IKONOS satellite images demonstrate the superiority of the proposed method in both spatial resolution improvement and spectral information preservation.
Get Citation
Copy Citation Text
Yin Wen, Li Yuanxiang, Zhou Zeming, Liu Shiqian. Remote Sensing Image Fusion Based on Sparse Representation[J]. Acta Optica Sinica, 2013, 33(4): 428003
Category: Remote Sensing and Sensors
Received: Sep. 5, 2012
Accepted: --
Published Online: Feb. 26, 2013
The Author Email: Wen Yin (yinwen@sjtu.edu.cn)