Research On Medical Image Fusion Algorithm Based On Detail Enhancement Decomposition

Fusion of multimodal medical images is an important branch of image fusion technology,which refers to the process of fusing two or more medical images from different image modalities into a single medical image by using a specific approach.The resulting medical fusion images are more complete and richer in terms of image information,texture structure and lesion characteristics than traditional medical fusion images from single image modalities,allowing clinicians to significantly improve clinical diagnosis and disease analysis.Typical problems encountered by existing medical image fusion algorithms are contrast degradation,blurred edge details and textures,color distortion,and time consumption.To address these problems,we have made the following main innovations:(1)The quality of the fused image depends on the feature extraction and fusion rules.The accurate separation of the basic layer and texture details is beneficial to achieve better results of the fusion rules,so it is especially important to extract the features adequately.Therefore,this paper proposes a detail enhancement decomposition model,which decomposes medical images into detail layers capturing rich gradient and energy information and basic layers containing energy information,and uses first-order derivative filters in two directions to enhance texture details and second-order Laplacian filters to smooth the decomposed layers.Extensive experiments prove that the method in this paper performs better on various medical images and achieves good quality in both detail enhancement and denoising.(2)On the basis of solving the above problems,the detail enhancement decomposition model is continued to be optimized.Firstly,since the decomposition of the detail enhancement model is defined on gradients,in order to cover meaningful layers more efficiently,this paper limits the scope of understanding to improve the computational efficiency of the detail enhancement model.Secondly,fusion rules are used for each layer.The product of the local L2 parametrization and the maximum singular value constitutes the fusion weight of the basic layer.The fusion weight of the detail layer is determined by the product of the local L1 parametrization and the maximum singular value.Finally,the fused image is reconstructed by fusing the basic layer and the fused detail layer.A large number of medical image fusion experiments confirm that this algorithm is more effective than many existing algorithms,with sharper edge details,better color and shorter running time.