The Research Of Multi-Modal Image Registration And Fusion

Compared to multi-modal image,the information cannot be completely obtained for mono-modal image.The method of image registration and image fusion are proposed and developed to date.The previous multi-modal image registration and fusion methods have achieved better effects,but there are still some problems.The defect that fall into the local optimum has been solved by biological intelligence algorithms,but single algorithm is lack of the fine tuning ability and is not suitable for large scale transformation.Many classical fusion algorithms are simple operation.However,anti-noise ability is poor for spatial domain fusion method;in terms of transform domain fusion algorithm,fusion effect is influenced by the decomposition classification and decomposition layers;intrinsic link between source images is not taken into account for traditional sparse representation algorithm.This paper did further research for the above two image processing problems.Algorithms are improved on the basis of predecessors and the improved algorithms are applied to the infrared and visible images.The main contents of this paper are as follows:Multimodal image registration algorithm is based on P system.Firstly,a cell-like P system of membrane structure is designed.Each object in membranes represents a group of transform parameters of floated images.Secondly,all objects of each elementary membrane is evolved by pseudo parallel differential evolution algorithm with dual subpopulations(DSPPDE).At the same time,the designed two exchange rules are used to update best parameters.Finally the global optimal object is stored in the skin membrane.Based on joint sparse model of clear edge image fusion method.Firstly,the source image is expressed as common sparse component and each individual sparse component with over-complete dictionary.Secondly,the designed fusion rule is acting on the common and innovation sparse coefficients to obtain the fused sparse coefficient.Finally,fused image is reconstructed from fused sparse coefficient and dictionary.