“…To test the superiority and feasibility of the algorithms, we chose the comparison methods from the various aspects, such as the methods focused on the image transformation, the methods based on sparse representation or filtering methods, the deep learning methods and so on. The comparison methods in this paper mainly include curvelet transform (CVT) [40], complex wavelet transform (CWT) [41], guided filtering fusion (GFF) [38], gradient transform (GTF) [42], hybrid multi-scale decomposition fusion (HMSD) [43], Laplacian pyramid with sparse representation (LP_SR) [44], ratio pyramid (RP) [45], fusion based on median filtering (TSF) [46], the weighted least square optimization-based method (WLS) [47], anisotropic diffusion fusion (ADF) [48], U2fusion [26] and Densefuse [25]. The implementation of these compared methods is publicly available, the parameters of which are strictly in accord with the original papers.…”