A combination of infrared intensity and polarization images can more fully describe the characteristics of a detected scene and facilitate subsequent processing. An algorithm for fusing infrared intensity and polarization images using hybrid l0l1 layer decomposition is proposed. The algorithm consists of the following steps. First, multi-scale geometric transformations are applied to the infrared polarization and intensity images using hybrid l0l1 layer decomposition. Then, in the low-frequency characteristic subband image, the index local Gaussian distribution similarity is adopted as the low-frequency image fusion weight of the infrared polarization image, and the fused infrared polarization image is injected into the low-frequency infrared intensity image. Next, the local spatial frequency and local energy are used to fuse the high-frequency subband image, and the two fused images are combined by principal component analysis to obtain a high-frequency fused image. The final fused image is obtained by reconstruction. An experimental comparison reveals that the algorithm can be used to fuse images of different types with complementary features, and the quality of the fused image is clearly improved.