基于改进LatLRR算法的红外与可见光图像融合研究OA北大核心CSTPCD

An improved latent low-rank representation(ILatLRR)is proposed to make the target more prominent and the background information more abundant after infrared and visible image fusion.First,the underlying layer obtained using LatLRR was decomposed at multiple levels to obtain additional underlying detail layers.Second,multilevel decomposition control was adopted based on the energy of the horizontal and vertical components of the detail layer and global contrast of the base layer to avoid invalid decomposition.Finally,different fusion strategies were adopted for the base and detail layers.The experimental simulations show that the fusion result of ILatLRR displays a sense of hierarchy;the image is clear;and the texture is rich.The contour details of the infrared thermal radiation target are maintained,retaining a large number of visible light image background features,with an objective evaluation index better than those of other algorithms.