A vision-based augmented reality system using an invisible marker indicates an invisible marker on a target object to be tracked, such that it can rapidly and correctly track the target object by detecting the invisible marker. The augmented reality system includes a target object including an infrared marker drawn by an invisible infrared light-emitting material; a visible-ray camera for capturing an image of the TO; an infrared-ray camera for capturing an image of the IM included in the TO image; an optical axis converter for allowing the infrared-ray camera and the visible-ray camera to have the same viewing point; an image processing system for rendering a prepared virtual image to the TO image to generate a new image.