Abstract: Although text-guided infrared-visible image fusion helps improve content understanding under extreme illumination, existing methods usually ignore semantic differences between textual and ...