Infrared And Visible Image Fusion Based On RPCA Model

Infrared and visible image fusion technology has became a hot issue in the field of image processing and machine vision with the continuous development of the imaging quality of the sensor,which can enhance the image quality and resolution.Visible light sensor imaging contains abundant details,which fits in with the observation of the human eyes,but affected easily by weather,can not work all-day.Though the infrared sensor has the characteristics of stable imaging,which can display the hidden target commendably and is less affected by lighting conditions and bad weather,it has the poor ability to reflect the details of the target due to the low contrast.Therefore the fusion of infrared and visible images is proposed to offset the shortcoming of both in order to give full play to their advantages.which make sure of the fusion image has the advantages of both infrared and visible images.In recent years,even though the infrared and visible image fusion technology has made great progress,the problems of which such as the distortion of the image,the lack of texture details and the significance of the target remain to be the key points and difficulties in the field of image fusion.In view of this situation,a method of infrared and visible image fusion based on RPCA decomposition model is presented in this paper as follows to solve this problem.1.Focus on the accurate registration of infrared and visible image in natural scene.This paper introduces the process,level,common methods and fusion rules of infrared and visible image fusion firstly;then discusses the process of image pre-processing;last but not least,the image registration technology in the process of image pre-processing is described,and the accuracy and stability of the infrared and visible image registration methods are verified by experiments and analysis.which provides a reliable basis for the pre-processing of infrared and visible image fusion.2.Focus on the feature description of infrared and visible images,this paper proposes the RPCA decomposition model of infrared and visible image by decomposing the infrared and visible images through RPCA on the basis of robust principal component analysis,and analyzes the feature information of the source image contained in the sparse matrix and the low rank matrix.3.Focus on the problem of image fusion method based on the non-subsampledContourlet transform such as fusion image distortion,texture detail information missing,target saliency.This paper proposes a novel infrared and visible image fusion method based on the RPCA decomposition model in NSCT domain.Firstly,the infrared and visible image are decomposed by RPCA to get the corresponding sparse matrix.Then,the infrared and visible images are decomposed by NSCT to obtain the low frequency sub-bands and high frequency sub-bands of the source image.For the decomposed image,the fusion method based on the sparse matrix is used to fuse.For the high frequency direction sub-bands,the maximum direction sub-bands is merged by the absolute value based on the sparse matrix,and the other layers are merged by PCNN method.Finally,the low frequency sub-bands and the high frequency sub-bands are reconstructed by NSCT to get the image fusion result.4.Experiments for standard gallery and real scene image set are carried out by using Contourlet transform fusion method,double NSCT fusion method,NSCTwavelet-PCNN fusion method and the method of this paper proposed.The experimental results show that comparing with the other fusion methods,the proposed method can not only highlight the target information in the infrared image,but also make the texture and detail information in the visible image more delicate,and can reduce the fusion image distortion.The evaluation results show that the method is more superior than another five objective evaluation indexes such as image mutual information,average gradient,standard deviation,peak signal-to-noise ratio and structural similarity index,which achieve an excellent performance.