Abstract. This study proposed a novel image fusion algorithm based on weighed multilevel morphological component analysis (FWMMCA). First, morphological component analysis is improved into a new multi-scale decomposition algorithm-multilevel morphological component analysis (MMCA). Then, feature vectors are extracted from MMCA sub images, which are used to reflect brightness, texture regularity, energy, and randomness of those images. Moreover, the feature vectors are innovatively used as weight in fusion rules, to propose a weighted fusion rules. Sub images are fused via those fusion rules, which are finally reconstituted into fused image.