Multi-Focus Image Fusion Algorithm Based On Rolling Guide Filter

Multi-focus image fusion is one of the focuses in the field of image fusion.Because of the limitation of the depth of the lens,the images which are focused within the depth of field but not outside the depth of field are obtained,multi-focus image fusion is used to make all objects in each focus area in the same scene appear clearly in the same image.At present,multi-focus image fusion technology has been widely used in digital photography,military,optical microscope,target detection and other fields.Rolling guidance filtering(RGF)is a kind of edge-preserving smoothing filter,which has the characteristics of small structure elimination,edge recovery and fast iterative operation.In order to solve the problems existing in multi-focus image fusion,the following two multi-focus image fusion methods based on RGF are proposed in this paper:(1)A multi-focus image fusion method based on RGF and CSR is proposed to solve the problems of blurred edges and lost details of focused objects.Since most traditional edge preserving filters smooth details according to contrast,spatial scale of images is seldom considered in the decomposition process.In order to achieve feature separation of images in the decomposition process,spatial scale should be considered,and RGF is just a kind of filter proposed based on spatial scale.Firstly,RGF and Gaussian filter are used to decompose the source images after registration.Secondly,for the base layer,the comparison saliency map and weight matrix are constructed for fusion.At the detail layer,the fusion rule of convolution sparse representation is used,mainly by solving the convolution sparse coefficients,to complete the fusion of characteristic response coefficients.Finally,the fusion result graph is obtained after reconstruction.Experimental results show that this method can avoid the edge blur of the target objects,and preserve the edges and details texture information of the source images well.(2)To solve the problem of low contrast of multi-focus fusion image,an image fusion method based on alternating gradient filter and improved pulse-coupled neural network(PCNN)is adopted.Combining rolling guided filter(RGF),smooth iterative recovery filter(SIRmed)and gradient filter(GF),a new filter,alternating gradient filter(AGF),is proposed,which can simultaneously achieve the characteristics of small structure elimination,local strength retention and edge recovery.Firstly,the multi-focus source images are decomposed into approximate layer and residual layer by alternating gradient filter.Secondly,for the approximate layer,the weight matrix is obtained by calculating its contrast significance diagram,and the regional energy of the weight is calculated to obtain the approximate layer fusion subgraph.Then,the residual layer is fused with the improved parameter adaptive PCNN fusion rule.Finally,the fusion result graph can be obtained by the reconstruction of alternating gradient filter.Through experiments and comparison with other fusion algorithms,it is verified that the proposed method can overcome the problems of poor contrast and uneven brightness distribution of fusion images,and better retain the details of source images.In this article,a multi-focus image fusion system based on rolling guide filter is designed,which can effectively analyze and evaluate the characteristics of the two methods.At the same time,seven image fusion methods and seven objective evaluation indicators are selected and compared with the method in this paper to highlight the advantages of the proposed method.The system also provides technical support for the application of multi-focus image fusion.