An image display device comprises a light-emitting unit (1), an optical fiber (2) which transmits light emitted from the light-emitting unit, and a plurality of spheres (3) disposed so as to be coupled to the optical fiber (2) by evanescent coupling and having optical transparency. Each of the spheres (3) produces an optical mode in which light entering each sphere from the optical fiber (2) by the evanescent coupling is confined inside, and are so formed that predetermined resonant frequencies of the spheres for producing the optical mode are different from one another, so that an image is displayed by the light emitted from the spheres (3). This provides a simply structured image display device having a high response speed and capable of achieving power saving, a high luminance, a wide view angle, and a high contrast.