An image display apparatus and method generate a computer-graphic image onto the display of a real scene, using output values from an attitude sensor to provide angle of inclination and fixed values for the angle of direction of the perspective of the display. The result is an image in which the CG image is aligned correctly along the angle of inclination and is misaligned only in the angle of direction. A user then adjust the angle of direction while observing the display and inputs an instruction when the virtual image and the real scene come into alignment, thus correcting the correction information. As a result, attitude adjustment is carried out more easily because it is not necessary for the user to fix the angle of inclination of the perspective.