Information Fusion And Enhancement Technology Research

Image fusion technology came into being in the initial military needs and achieved continuous development under the circumstance of informational internet plus era.However,the traditional fusion methods still have problems such as difficulty in distinguishing military camouflage targets,poor imaging quality in low light situations,inability to obtain target details and poor fault tolerance.Therefore,this paper focuses on the feature-based image fusion algorithm,using the polarization information of the targets to establish the targets' polarization property model.In this paper,the target images are fused from the perspective of feature-based fusion to achieve the effect of image enhancement.By studying the polarization state of light,many expressions describing the influence of objects on the polarization state of incident light are obtained,in which the expression of the polarization state by the Mueller matrix is the most comprehensive.Through the analysis of the Mueller matrix,various parameters describing the polarization state of the exiting light and the polarization characteristics of the objects can be obtained,which lays the foundation for establishing the polarization property model of the targets.By combining the deformation of the Mueller matrix with the Fresnel formula,the refractive index and extinction coefficient expressed by the Mueller matrix elements can be deduced.The addition of these two parameters makes the polarization property model more comprehensive.According to the principle of time-sharing measurement of Mueller matrix,a convenient and accurate measurement method and a set of multifunctional experimental apparatus are designed.According to the actual measurement requirements,the optical path of Mueller matrix measurement is set up in order to obtain accurate Mueller matrix images.According to the polarization property model,the target images are fused with the common fusion algorithms at the feature level.The results show that the information entropy,the average gradient and the mutual information of the fused images obtained by establishing the polarization property model are significantly improved,and the position,contour and detail of the target are effectively enhanced.