Research On The Fusion Of Infrared And Visible Images Based On The Curvelet Transform

In recent years, the fusion of infrared and visible images becomes a researchhotspot in the field of image processing. Because the infrared and visible images havedifferent imaging characteristics and good complementary between the imagesinformation, the integration of the two kinds of images can reflect the targetcharacteristics of the infrared image and the scene information of the visible imagemore effectively, making it have good prospects in the military and civilian fields.Image fusion contains two important factors: the multi-scale image decompositionand reconstruction and the fusion rules of decomposition coefficient. In a variety ofmulti-scale decomposition methods, wavelet transform has undoubtedly made a greatsuccess, mainly because it has a good time-frequency locality, which can effectivelyreflect the point singularity. It is difficult to effectively deal with the line singularity(edge) in two-dimensional image. In order to overcome the defects of the wavelettransform analysis in two-dimensional or higher-dimensional spatial, the Curvelettransform used in image fusion can effectively analyze curve singularity of the image,which can deal with the edge information of the image more reasonable and improve thequality of integration image. Therefore, this paper presents an improved image fusionalgorithm based on the second generation Curvelet transform, and formulates the fusionrules of low-frequency and high-frequency coefficients respectively for differentimaging characteristics of the infrared and visible images. And the experimentsachieved good results in the fusions of two sets of source images.In addition, a fusion strategy based on the HSI transform is proposed for theproblem of infrared and color visible images fusion. Firstly, the RGB image is mappedto the HSI space. Then the gray component I and infrared image are fused based on theCurvelet transform. Finally the combination of the fused image used to be a new graycomponent and the components H and S of source color image is mapped to RGB colorspace, resulting in the fusion image. Through comparative analysis of fusionexperiments, good result is also received.