This paper reports the improvement of image quality issue in the fusion of remote sensing images by minimizing an energy function. By gradient constraint term of the energy function, the spatial information of the panchromatic imagewastransferredtofusedresults.Thespectralinformation of multispectral image was preserved by importing a kernel function to the data fitting term of the energy function. Visual perception measurement and selected fusion metrics were employed to evaluate the fusion performance. Experimental results demonstrated the proposed method outperforms other established image fusion techniques.