Multi-spectrum Image Fusion Based On Image Enhancement

Multi-band image fusion is an important way of information representation,which has important applications in many fields such as medical,military,and security.In multi-band image fusion,each band image captures information and synthesizes it through corresponding sensors.In the capture of each band image,there are many degradation factors affecting the imaging quality,including environmental conditions,limitations of the optical system,noise of the acquisition system,etc.These factors lead to the degradation of image quality and subsequent image fusion.In addition,in the process of image fusion,different image fusion methods will also affect the quality of image fusion.If the effective information of each band cannot be extracted in multi-band image fusion,it may affect not only the information efficiency in the fusion but also more noise,which will affect subsequent processing and discrimination.In this paper,the image restoration enhancement technology is used to restored and enhance the long-wave infrared image,middle-wave infrared image and visible image.The noise in each band image is removed by various means to improve the image quality.At the same time,the significant information in each band image is extracted and multi-scale analysis is performed based on this.The band image is fused to improve the information integration efficiency in the image fusion process.By observing the expression of infrared non-uniform fringe noise,the distribution in mathematical statistics is analyzed.As the non-uniformity stripe noise is generally distributed in a single direction in the infrared image,an infrared non-uniformity stripe noise suppression method with the horizontal gradient minimization and the vertical gradient remaining unchanged as the core is proposed.The approach reduces the non-uniformity stripe noise in both long-wave infrared image and middle-wave infrared image as much as possible.Considering the effect of the optical system on image degradation in the visible light band image,images at different fields of view of the same optical system have different degradation models.The degradation model of different fields of view can be obtained by analyzing the MTF curves of different fields of view of the image,and the images in different fields of view of the optical system are restored by using different degradation models to ensure that each field of view is in a reasonable state when it is restored.Since the final meaning of dual-band image fusion is the synthesis of effective information in each band image,it is proposed to use the local saliency method to extract the feature information in each band and to target the dual-band image fusion.On the basis of this,the method of guided filtering is used to multi-scale the image to smooth the dual-band image at each scale,avoiding the image mutation of adjacent regions in the fused image.