Image fusion by combining SWT and variational model

Abstract

A new model is presented to perform image fusion by combining discrete stationary wavelet transform (SWT) and the variational approach. After the input images are decomposed with SWT, the approximate and detail coefficients of the fused image are chosen with the variational model considering the fusion requirements such as the spatial quality improvement or structural similarity preserving. The gradient descent flow is applied for the optimization solution of the variational model. The method is compared with the wavelet-based, SWT-based and contourlet-based fusion models visually and quantitatively. Experimental results demonstrate that this model can achieve the prospective results.