A Hybrid Approach for Fusion Combining SWT and Sparse Representation in Multispectral Images

Abstract

Background/Objectives: Image fusion in remote sensing is a challenging task for fusing minute differences in multispectral images for further analysis. The objective of this paper is to propose a hybrid approach for image fusion in remotely sensed images. Methods/Statistical Analysis: The objective of this paper is accomplished by a hybrid approach combining Stationary Wavelet Transform (SWT) and sparse representation. SWT is used for pre-processing and sparse is used for fusing the multitemporal LANDSAT image. Results: Various fusion metrics like RMSE, PSNR and FMI are evaluated and the obtained results show that the proposed hybrid approach outperforms well than the existing methods. The proposed approach gives better results in terms of all the features has been fused correctly, less mean square error and high signal to noise ratio compared to the existing methods such as DWT, SWT and Ehlers. Conclusion/Application: The application of this work is mainly on multispectral multitemporal images. It can be used for change detection also can be used to fuse low resolution image with high resolution remote sensing image. Finally this approach provides an efficient fusion result compared to the traditional methods.