Research On Multi Band Infrared Image Fusion Algorithm

With the development of high technology,precision guided weapons have become the main weapons of modern warfare.And the research of image fusion technology in imaging guidance technology is also one of the focuses of precision guidance technology nowadays.The fusion of multi-source image information of the same scene can synthesize the complementary information of different sensors,making the fusion of the image content more abundant,the details more clear,can make operational weapons adapt to the complex and changeable battlefield environment(such as the target stealth,etc.).In addition,multi-sensor image fusion technology has also been widely used in the civilian field,such as intelligent robots,remote sensing,medicine,and manufacturing.From the late 1970 s to the present,image fusion technology has been developed for many years,but the fusion technology of mid and long wave images in pure infrared band still has much room for development.Therefore,according to the characteristics of infrared multiband images,it is very meaningful and necessary to study the image fusion algorithm pertinent.Non-subsampled contourlet transform(NSCT)is another very popular image analysis method after wavelet transform in recent years.It not only has the multi-resolution analysis characteristics of wavelet transform,but also has the properties of higher directional and anisotropic that wavelet transform does not have.It is a better image analysis method.Especially when there are more lines and contours in the image,the advantage of NSCT is more obvious.This paper focuses on NSCT,we have studied the whole process of image fusion from infrared multi band image acquisition,infrared multi band image registration to the final image fusion.Firstly,the target characteristics of infrared medium-wave image and infrared long-wave image are analyzed.Then,according to the characteristics of infrared images,a pre-processing method for image fusion such as de-surfacing and registration,which is more suitable for infrared images,is designed.Through experiments,it is verified that the effect of fusion after pre-processing is better than that of direct fusion.Then,non-subsampled contourlet transform theory,non-negative matrix factorization theory and sparse representation theory are mainly analyzed.According to the target characteristics of infrared medium-wave images and long-wave images,Combines non-subsampled contourlet transforms with non-negative matrix factorization,an image fusion algorithm based on non-subsampled contourlet transform and non-negative matrix factorization,which is more suitable for infrared images,is proposed.Then combining non-subsampled contourlet transform and sparse representation,an image fusion algorithm based on non-subsampled contourlet transform and sparse representation is proposed,which is more suitable for infrared images.Finally,compared with the conventional non-subsampled contourlet transform and other traditional image fusion algorithms based on wavelet transform,the effectiveness of the proposed algorithm in infrared multi-band image fusion is proved.