Research Of Multi-source Image Fusion Methods In Frequency Domain

With the rapid development of information technology,the diversity of data sources,and the sharp increase in information volume,people put higher and higher requirements on information's,especially on images',comprehensiveness and accuracy.Information contained in the images from different sensors is similar but diverse,due to the different types and functions of those sensors.Correspondingly comes the Multi source image fusion technology.This technology becomes a useful and sharp weapon of researchers.In this paper we focus on how to integrate the images acquired by different sensors,so that the new image contains more favorable information of the source image.Multi source image fusion technology combines two aspects of spatial domain and frequency domain.The frequency domain method,due to its characteristics of multi-scale and multi-resolution and a more precise expression to the details of the image,is more welcome by the researchers and has become the mainstream method of multi-source image fusion.This paper,based on natural multi focus and medical diagnosis,studies and creatively improves the relevant algorithms.Since multi-focus image fusion requires high-resolution results,we choose the VCT transform,a combination of V-transform and Contourlet transform,and non-subsampled Contourlet transform(NSCT)method to decompose the source images.We select the statistical features that favor the expression of image clarity to fuse the transformed coefficients and obtain the fxsed image by inverse transformation.For medical images,due to their requirement on relatively high precision,we choose V-multilayer decomposition.to divide the source images into two parts,contours and details,fuse them respectively,and,sum up the contour information and the details to obtain the final fusion image.This is the first time to apply the V-system,which is good at describing the details and the discontinuous points,to image fusion,and add in the popular Contourlet transform and NSCT transform,allowing the images to obtain a multi-level,multi scale and multi direction decomposition and lay a solid foundation for subsequent fusion.This paper also applies the edge detection algorithm to improve the quality of initial fusion image,so that the final fusion image is closer to the source image and reflects more favorable information.A large number of experimental data show that the proposed method performs well in both subjective and objective data,which proves the effectiveness of the method.