A Study Of Image Fusion Algorithms Based On Adaptive Guided Filter

Image fusion is the combination of images from multi-sensors to obtain fused image with more information.Image fusion technology has a good application prospect in medicine,remote sensing,land use management and objects recognition.It also has wide applications in disease diagnosis and improvement of the performance of imaging system.According to the characteristics of the guided filter and the purpose of fusion of three types images,image fusion algorithms based on adaptive guided filter are proposed.The specific contents are as follows:A CT and MRI image fusion algorithm based on adaptive guided filter is proposed.Standard deviation,average gradient and edge strength are used as evaluation parameters and objective functions.The parameters of the guided filter are obtained by the optimization method.The high and low coefficients of CT and MRI image are obtained respectively with multi-scale transform.The refined weight maps are obtained after guided filter of weight map.The fused coefficients are obtained by weighted fusion method with refined map as weight value.Fused image is obtained by inverse multi-scale transform of fused coefficients.The proposed algorithm is compared with Choose-max and intuitionistic fuzzy algorithms by using 15 groups of CT and MRI experimental images.Experimental results show that the proposed algorithm is better than other two algorithms in terms of standard deviation,average gradient and edge strength.Subjective observation show that the fused image obtained by the proposed algorithm contains more clear bone tissue and soft tissue that can be used in medical diagnosis,lesion target location.A MS and PAN image fusion algorithm based on adaptive guided filter is proposed.Average gradient and the spatial frequency are used as the evaluation parameters and objective functions.The parameters of the guided filter are obtained by the optimization method.Three components of I,H and S are obtained by IHS transform of MS image.The fused I component is obtained by weighted fusion method with guided filter.Fused image is obtained by inverse IHS transform for fused I component and the original H and S components.The proposed algorithm is compared with Choose-max and intuitionistic fuzzy algorithms by using 12 groups of MS and PAN experimental images.The experimental results show that the fused image obtained by this algorithm is better than that of the Choose-max and the intuitionistic fuzzy algorithms.So this algorithm is better than other two algorithms.A multi-focus image fusion algorithm based on adaptive guided filtering is proposed.Standard deviation and average gradient are used as evaluation parameters and objective functions.The parameters of the guided filter are obtained by the optimization method.High and low frequency coefficients are obtained by multi-scale transform of multi-focus images.The fused high and low coefficients are obtained by fusion method with guided filter respectively.Fused image is obtained by inverse multi-scale transform of fused coefficients.The proposed algorithm is compared with Choose-max and intuitionistic fuzzy algorithms by using 12 groups of multi-focus experimental images.Experimental results show that this algorithm is better than Choose-max and intuitionistic fuzzy algorithms.Adaptive guided filter image fusion algorithms proposed in this thesis can better fuse CT and MRI images,MS and PAN images,multi-focus images respectively,and can obtain better image fusion results.Algorithms proposed may be widely used in many application fields.