| AD is a major disease threatening the health of the elderly.With the development of imaging,different modal imaging has become an important basis for doctors’ clinical diagnosis.The single modality reflects the limited information of human brain tissue.PET image reflects the activity of nerve synapses physiologically yet with low resolution,while MRI image shows clear brain tissue structure yet can’t be traced back to the pathologica.Combining the advantages of PET and MRI images to generate fusion images can improve the specificity and accuracy of AD diagnosis.Currently,the ShearLab 3D transform in frequency domain multi-scale transform method still has the problem of blurred edge details.The structure tensor in spatial domain multi-scale transform method still has the problem of unstable quality and low contrast.In view of the different edge and texture characteristics of the high frequency subbands of ShearLab 3D transform,this paper proposes a three-band decomposition combining the ShearLab 3D transform and variance threshold.The whale optimization algorithm is introduced to obtain the optimal variance threshold,and the high frequency subbands are divided into medium and high and high frequency subbands containing edge features and detailed features.Considering that the smooth weight matrix of low frequency fusion rules will weaken the detailed texture information of the image,an improved weight matrix for liveness measurement is proposed.In view of the influence of ShearLab 3D transformation parameters on the fusion result,this paper proposes to determine the optimal scaling and shearing parameters based on the subjective and objective evaluation indexes of the fusion results.Aiming at the problem of unstable image quality and low contrast in the fusion image in structure tensor algorithm,this paper proposes a image fusion method based on the combination of improved structure tensor and whale optimization algorithm.Aiming at the prominent bright spots in fusion images,this paper proposes to decompose PET and MRI images by adopting different trace thresholds.Aiming at the problem of low contrast of the fused image,a linear enhancement is proposed by giving different weights to the smooth area and edge area.Considering that trace threshold and weight value only determined by experience will lead to unstable fusion results,a whale optimization algorithm is proposed to determine the optimal combination of parameter values.The experimental results show that the subjective and objective evaluation of three-band decomposition based on the combination of ShearLab 3D transform and the variance threshold outperforms the two-band decomposition.The low frequency fusion rule based on the improved activity measurement weight matrix is superior to other weight matrix.The scale and shearing parameters of ShearLab 3D transformation are determined through experiments.The parameter optimization of improved structure tensor based on whale optimization algorithm outperforms others on subjective and objective evaluation.The improved structure tensor combining whale optimization algorithm yields better results than other structure tensor. |
