Image fusion method based on a new class of wavelet non-separable wavelet with compactly supported, symmetric, orthogonal and dilation matrix is presented. We first present a new constructing method of two dimensional 4x4 symmetric filter bank. Using this method, we construct a non-separable wavelet 4-channels filter bank. The images involved the fusion were decomposed by the filter bank. We use the fusion algorithm as follow: for low-frequency part, the average value was selected, and selecting the center pixel value of the local area window whose the energy is bigger for the three high-frequency parts of each level, then the new image was reconstructed. The method's performance was evaluated by using the entropy, cross-entropy, mutual information, error of mean square root and points signal to noise ratio. The experiment results show that it has good effect for the fusion of visible image and infrared image, far infrared image and near infrared image, aero image and satellite image, multi-focus images and many others class of images.We can obtain the fused image whose edges is no distortion.The performance of the non-separable wavelet fusion method is better than the performance of the haar separable wavelet fusion method.