Research On Infrared Dual-band Image Fusion Technology

With the development of infrared detectors and image processing technology,various infrared detector systems are widely used in military and civilian fields,and are developing in the direction of high resolution,multi-band and large area array.Among them,the infrared dual-band detector has developed rapidly,called the third-generation infrared detection system.The information of different bands is obtained,the information difference between the bands is analyzed and processed,and the complete information expression of the same scene is obtained finally.At the same time,image fusion technology can combine the images obtained by multiple sensors with effective information.It is widely used in information fusion of various types of images,such as infrared images,multi-focus images,various medical images,remote sensing images,micro light image,and polarized image,etc.In this paper,the related technologies of infrared dual-band image fusion algorithm are studied.And the Long Wave Infrared(LWIR)and Medium Wave Infrared(MWIR)images are taken as research objects,the main research contents are as follows:1.The characteristics of LWIR and MWIR images were studied.The infrared images of the two groups are analyzed from the perspectives of the difference of the digital statistical features,the difference of the target and the background,and the difference of the image transformation domain.Two kinds of infrared images suitable for image fusion are obtained.One is the infrared picture group applied to the urban traffic monitoring system.The infrared target in the picture is complex,the background information is more,and the edge and texture are richer.The fusion can be realized by the traditional image fusion algorithm.The other is the infrared long-distance target picture group.The infrared target in the picture is small,and the background noise and clutter are more.Since the traditional fusion algorithm cannot distinguish its background and target,it is necessary to introduce the target detection algorithm into the fusion algorithm of the infrared picture group.2.An image fusion algorithm based on Visual Saliency Map(VSM)and Adaptive Pulse Coupled Neural Network(APCNN)is proposed.Aiming to solve the problems of low brightness and poor contrast of infrared images,VSM-based low-frequency fusion rules and APCNN-based high-frequency fusion rules are adopted.In this paper,the fusion rulesare improved respectively.The energy of image gradient(EOG)is introduced in the low-frequency fusion to extract the remaining high-frequency information in the low-frequency information,and the spatial frequency(SF)is used in the high-frequency fusion.Improve link strength in APCNN.By comparing with five traditional fusion algorithms,the effectiveness of the algorithm is illustrated from the aspects of subjective visual effects and seven objective evaluation factors.3.An infrared enhanced dual-band target enhancement fusion algorithm is proposed.Since the small target accounts for a small proportion in the MWIR and LWIR images,it is easy to introduce a large amount of background noise,which seriously affects the fusion effect.Therefore,this paper suppresses the background noise by enhancing the target area by the two steps of initial image fusion and target enhancement fusion in the initial fusion.In order to reduce the background noise and the energy of the clutter,the low frequency sub-band uses the weighted fusion rule to reduce the background.The low-frequency sub-bands with more clutters fuse the weights to achieve the effect of suppressing the background.In the target-enhanced fusion,the morphological filtering and iterative thresholds are used to extract the potential target regions,and the potential target regions of MWIR and LWIR are superimposed and then non-existent.Linear enhancement to achieve the goal enhancement effect.The experimental results show that the obtained target enhanced fused image has a great improvement in visual effects such as contrast and brightness,and also has a good improvement in terms of signal-to-noise ratio and signal-to-noise ratio.