Research On Image Fusion Based On Guide Filter In NSCT Domain

As the rapid development of information fusion technology,image fusion as an important branch,has also received extensive attention.Because of the different imaging mechanism.infrared sensor and visible light sensor have their own merits and faults.The infrared images are stable less affected by environment,Meanwhile,they can reflect the target information better.However,because their imaging principles are not consistent with human vision habit,it is difficult to observe and analyze directly.The visible light images accord with the observation habit of human eyes,they reflect the degree of detail information,but The imaging effect usually depends on the external conditions.So complementing one another by fused the infrared and visible light images.The fusion images can show background information and target information provided by visible images and infrared images separately.It is more suitable for observation and later study.The two key problems are how to improve the fusion effect while reducing the calculation in the process of fusion.To make up for defects,In this paper,we propose a fusion algorithm with more prominent targets,more abundant information and more timeliness:Infrared and visible light images fusion based on Target extraction and Guide filter in FNSCT domain.The main contributions are as follows:1.Analysis of the characteristics of infrared images under natural conditions.before multiscale decomposition,extract Infrared targets based on regional significance and K means clustering to get infrared target and background images.Experimental results show that this method can extract the Infrared targets effectively and accurately.2.Analysis of the principle of image multiscale decomposition methods and the basic theories of image fusion,traditional NSCT decomposition is time-consuming.To solve the problem of timeliness,this paper presents a fast realization algorithm(FNSCT)based on NSCT.In order to verify the effectiveness of FNSCT,we compare the processing time including decomposition processing and reconstruction processing of different sizes images.Experimental results show that FNSCT is efficient and robust.3.Focus on the defects of low edge retention rate in exist fused images,we use Guide filter to modify the initial fusion weights for low frequency and high-frequency parts.The proposed algorithm can enhance the edge retention and reduce computation ofthe fused images at the same time.Experimental results show that it is an efficient algorithm to fuse infrared and visible images.With the proposed method,the targets is more prominent,the background is clearer in the fused images.The proposed method is robust and All the indicators has good performance in various indicators.