Abstract: A new image fusion method based on Adaptive Guided Filtering (AGF) and Principal Component Analysis (PCA) is proposed to address the issues of poor adaptability, insufficient fusion, feature attenuation, and detail loss in current infrared and visible light image fusion methods in different scenarios. Firstly, the visible light image is preprocessed through parameter adaptive guided filtering. In the layering stage, AGF is used to obtain the base layer representing the main body contour and the detail layer representing the texture details. In the fusion stage, based on the PCA feature space projection theory, the directional extraction and fusion of main features are achieved by analyzing the eigenvectors of the covariance matrix, and then the maximum absolute value method is used to jointly complete the fusion of sub layers. The experimental results show that the standard deviation of structural information can be improved by up to 42.5% compared to other fusion algorithms. The visual effect not only significantly highlights the target features but also fully preserves the details and textures, demonstrating better target recognition and scene adaptability.