Medical image fusion based on quasi-cross bilateral filtering

Image fusion technology is a method that uses a specific algorithm to extract and fuse features from multiple images and then combine them into one image. Unfortunately, most of the current image fusion methods can't balance the fusion performance, time consumption and model complexity. Firstly, in the classical bilateral filtering algorithm, the pixel domain originally had only the pixel value of a single image, and the unique information of the image relative to another source image can not be obtained, resulting in partial loss of the lesion information in the fused image. Based on the above problems, an image decomposition method based on quasi-cross bilateral filtering (QBF) is proposed. The image is decomposed into an energy layer with pure intensity information and a structure layer with rich details. Then, the visual saliency detection map (VSDM) is used to guide the fusion of the energy layers to take full advantage of the edge contour extraction to improve the edge contour sharpness of the fused image and to retain the image detail information fully. The improved multi-level morphology gradient (IMLMG) combined with the weighted sum of eight-neighborhood-based modified Laplacian (WSEML) was used to guide the fusion of the structure layer, which effectively improved the sharpness of the contour edges of the fused image. Finally, to evaluate the performance of the proposed fusion method, we fused six groups of medical images using the proposed method, and then evaluated the fusion quality by subjective vision and objective quality evaluation. The results showed that the proposed method performed better than nine other methods in terms of providing detailed information, edge contour, and overall contrast.