Weighted image fusion based on multi-scale top-hat transform: Algorithms and a comparison study

Abstract

Image fusion is an important technique in image processing. The key of image fusion is importing all the possible useful image features into the fusion result image. Top-hat transform could extract bright and dim image features. Multi-scale top-hat transform could extract all the image features at different image scales. Therefore, the multi-scale top-hat transform could be used for image fusion. In this paper, the weighted image fusion algorithms based on the multi-scale top-hat transform are discussed. The multi-scale top-hat transform using multi-scale structuring elements with the same shape and increasing sizes is used to extract the useful bright and dim image features for image fusion. To appropriately import the extracted useful bright and dim image features into the fusion result image, three weighted image fusion strategies are discussed, which gives three effective algorithms for image fusion. Moreover, the visual and quantitative comparisons of these algorithms with some widely used algorithms are also discussed. And, the experimental results show that, the weighted image fusion algorithms based on the multi-scale top-hat transform could achieve satisfying results in all the visual and quantitative comparisons. And, the results suggested that, because the mean value weighted image fusion algorithm obtains clearer result images with more image details than other algorithms and achieves the best overall performance on all the measures, the mean value weighted image fusion algorithm could achieve a better performance. All of the experimental results indicate that, the weighted image fusion algorithms based on multi-scale top-hat transform may be widely used in different applications, such as target detection, object recognition and security surveillance, for image fusion.