Research Of Image Fusion Algorithm Based On Contourlet Transform

Image fusion technology is widely used in the fields of remote sensing,military,medical,etc.It is becoming a promising paradigm for research.In order to get fusion methods with good effect,basic theory of Contourlet transform and the relationship between Contourlet coefficient are discussed in the paper.Then,three kinds of Contourlet correlation coefficient are proposed,and neighboring-maxima child node correlation coefficients is applied to the image fusion algorithm,and then an image fusion algorithm based on Contourlet correlation coefficient is proposed.Pulse Coupled Neural Network(PCNN)is an intelligent neural network,similar to the human visual system,is widely used in image processing.Applying PCNN in image fusion,their combination is accord with visual characteristics.In this paper,an image fusion algorithm based on Contourlet correlation coefficient combining with pulse coupled neural network is proposed.In this paper,we study the contents above and the research contents are as follows:Firstly,Contourlet transform,which has characteristics of multi-scale and multi-directional,is studied.Contourlet Transform has two components : Laplace pyramids and directional filter banks;In order to creat a basis for analysis of the correlation coefficients,the feature of Contourlet transform is introduced;According to transmission of Contourlet decomposition coefficients,the relationship between the inner-scale and the relationship between the inter-scale of Contourlet coefficients are discussed.Then,3 Contourlet correlation coefficients are proposed.The effects of edge features and textures extracted respectively with the aid of 3 correlation coefficients are compared by the experiments.Secondly,the basic theory of pulse coupled neural networks is introduced;Thebasic model of PCNN is introduced,and the PCNN composition is introduced in detail,the relationship between various parameters in PCNN is studied;then operating mode of no-coupling link PCNN and coupling link PCNN are introduced,respectily;a simplified model of PCNN is introduced.Thirdly,an image fusion algorithm based on Contourlet correlation coefficient is proposed.For the low-frequency sub-band coefficients and high-frequency directional subband coefficients,different fusion rules are selected.Neighboring-maxima child node correlation coefficients combined with energy weight is used as the fusion rule for high-frequency decomposition coefficients.Low-frequency coefficients are fused by the method of weighted.Experimental results show that the fused image efficiently represents the edge and texture information.Fourthly,combining Contourlet correlation coefficient with simplified PCNN model,an image fusion algorithm is designed,the specific steps are: after the original images are decomposed by Contourlet transform,low-frequency coefficients and high-frequency directional sub-bands coefficients are obtained.Two different rules are used.For the low-frequency sub-band coefficients,weighted average fusion rule is selected;For the high-frequency sub-band coefficients,Contourlet correlation coefficient Contourlet is selected as PCNN input,while the region sharpness is selected as the value of the parameter link strength ? in PCNN.Experimental results show that the fusion image not only ensure the fusion effect but also accord with human visual characteristics.